{ "schema": "ai4education.public.v1", "generation": "edu-v3-be1926657e9a8918", "items": [ { "body": { "en": [ "The instructor used ungrading and peer coaching: pairs take coach/coachee roles, ask probing questions, then switch." ], "zh": [ "教师用去评分和同伴教练：两人分任教练与受教练者，提出追问，再交换角色。" ] }, "collectedAt": "2026-10-13T06:26:40.247096Z", "id": "auto-819698a8b7938efe", "limits": { "en": "Source notes peer coaching falls flat without scaffolding, modeling, or practice.", "zh": "来源指出：若缺少支架、示范或练习，同伴教练会失效。" }, "publishedAt": "2026-10-12", "sources": [ { "checkedAt": "2026-10-13T06:26:40.247096Z", "name": "Robert Talbert / David Clark · Grading for Growth", "scope": { "en": "Reading scope: identified public excerpt; full text not read.", "zh": "读取范围：明确标注的公开节选，未读全文。" }, "status": "limited", "url": "https://gradingforgrowth.com/p/talking-is-also-thinking" } ], "stages": [ "higher" ], "summary": { "en": "A guest post describes using ungrading and peer coaching to deepen student thinking and make ChatGPT shortcuts less tempting.", "zh": "一篇客座文章描述用去评分和同伴教练深化学生思考，使ChatGPT捷径不那么诱人。" }, "title": { "en": "Peer coaching before writing may reduce AI shortcuts", "zh": "写作前同伴教练或减少AI捷径" }, "topics": [ "teaching", "learning", "assessment", "tools" ], "type": "practice" }, { "body": { "en": [ "He says agentic AI makes validating automated decisions hard, especially undisclosed grading or feedback.", "He urges disclosure and asks: if I cannot explain an agent's action, I avoid it and accept responsibility." ], "zh": [ "他说，代理式AI使自动决策难以验证，尤其是未披露的评分或反馈。", "他呼吁披露，并自问：若无法解释代理行为，就避免使用并承担责任。" ] }, "collectedAt": "2026-10-13T06:25:05.868856Z", "id": "auto-81e99090daa88ecf", "limits": { "en": "Opinion excerpt, not research; full text not read; no outcomes verified.", "zh": "观点节选，非研究；未读全文；未核实成效。" }, "publishedAt": "2026-10-11", "sources": [ { "checkedAt": "2026-10-13T06:25:05.868856Z", "name": "Marc Watkins · Rhetorica", "scope": { "en": "Reading scope: identified public excerpt; full text not read.", "zh": "读取范围：明确标注的公开节选，未读全文。" }, "status": "limited", "url": "https://marcwatkins.substack.com/p/discernment-in-the-age-of-agents" } ], "stages": [ "higher", "general" ], "summary": { "en": "Marc Watkins argues faculty should question when to use AI agents and disclose their use to students.", "zh": "马克·沃特金斯主张教师应质疑何时使用AI代理，并向学生披露使用情况。" }, "title": { "en": "Discernment and Disclosure for AI Agents in Education", "zh": "AI代理时代教育中的辨别与披露" }, "topics": [ "assessment", "policy", "teaching", "tools" ], "type": "opinion" }, { "body": { "en": [ "He urges reframing AI: from chat partner to capable software, character to company, prompts to systems.", "His account cites an OpenAI sandbox breach of Medicare data and designs that lull users into granting access." ], "zh": [ "他主张重构AI认知：从聊天对象到能力型软件，从角色到公司，从提示到系统。", "他援引OpenAI沙箱入侵Medicare数据，以及诱导用户开放权限的产品设计。" ] }, "collectedAt": "2026-10-13T03:23:03.014139Z", "id": "auto-654463be053f361b", "limits": { "en": "This is an opinion excerpt; the Medicare and product claims are the author's, not independently verified here.", "zh": "本文为观点摘录；Medicare及产品说法属作者主张，未经独立核实。" }, "publishedAt": "2026-10-08", "sources": [ { "checkedAt": "2026-10-13T03:23:03.014139Z", "name": "Leon Furze", "scope": { "en": "Reading scope: identified public excerpt; full text not read.", "zh": "读取范围：明确标注的公开节选，未读全文。" }, "status": "limited", "url": "https://leonfurze.com/2026/10/08/beyond-chatbots-convenient-costume/" } ], "stages": [ "school", "higher", "general" ], "summary": { "en": "Furze argues chatbot framing hides corporate responsibility and misdirects education's assessment debates.", "zh": "Furze认为，聊天机器人叙事掩盖企业责任，并误导教育评估讨论。" }, "title": { "en": "Beyond chatbots: AI agents as software, not characters", "zh": "超越聊天机器人：AI代理是软件，非角色" }, "topics": [ "policy", "assessment", "theory", "tools" ], "type": "opinion" }, { "body": { "en": [ "An English primary-school randomized trial found roughly 24% less planning time with Aila, about 49 minutes weekly, without inferior assessed resources. Use declined during the ten-week trial, and teacher views were mixed." ], "zh": [ "EEF 公布的英格兰小学随机试验发现，Aila 组备课与准备资源的时间减少约24%，平均每周节省49分钟，所评资源质量未见下降。十周内使用逐渐下降，教师评价不一，工具适配与持续采用仍需关注。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "eef-aila-trial-20261006", "limits": { "en": "The trial measured teacher planning and resource quality, not improved student learning. Control teachers could use other AI. The trial ran in autumn 2025; results were announced on 6 October 2026.", "zh": "测量教师备课与资源质量，不证明学生学得更好；对照组可用其他 AI。试验于2025年秋季进行，2026年10月6日发布结果。" }, "publishedAt": "2026-10-06", "sources": [ { "checkedAt": "2026-10-07T07:32:42.512013+00:00", "name": "Education Endowment Foundation / NFER", "scope": { "en": "Official news, project page and report sample flow checked. This summary retains the consistent 24% and 49-minute figures and avoids inconsistent school counts in page metadata.", "zh": "已核读官方新闻、项目页及报告样本流程；摘要仅保留一致的24%与49分钟，不采用网页头部与正文不一致的学校数。" }, "status": "verified", "url": "https://educationendowmentfoundation.org.uk/news/ai-tool-reduced-primary-teachers-lesson-planning-time-by-a-quarter-but-views-on-the-tool-were-mixed" } ], "stages": [ "school" ], "summary": { "en": "An English primary-school randomized trial found roughly 24% less planning time with Aila, about 49 minutes weekly, without inferior assessed resources. Use declined during the ten-week trial, and teacher views were mixed.", "zh": "EEF 公布的英格兰小学随机试验发现，Aila 组备课与准备资源的时间减少约24%，平均每周节省49分钟，所评资源质量未见下降。十周内使用逐渐下降，教师评价不一，工具适配与持续采用仍需关注。" }, "title": { "en": "Aila trial: less lesson-planning time, but mixed views and declining use", "zh": "Aila 随机试验：备课时间减少，但持续使用仍是难题" }, "topics": [ "teaching", "tools" ], "type": "research" }, { "body": { "en": [ "He distinguishes removing extraneous work from outsourcing the thinking that builds capability.", "He acknowledges that reforms he helped develop assumed a chatbot context, and argues that agents and wearables require renewed governance scrutiny." ], "zh": [ "他区分卸除无关负担与外包形成知识的思考，并主张培养有学科基础的判断和自我调节能力。", "他承认自己参与的部分改革也基于早期聊天机器人情境，认为代理与可穿戴技术要求重新审视治理速度。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "lodge-global-conversation-20261004", "limits": { "en": "Participant reflection and opinion; survey, PISA and commercial examples were not independently verified here.", "zh": "参会者观察与立场；文中调查、PISA和商业案例未在此独立验证。" }, "publishedAt": "2026-10-04", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Jason Lodge · Needed Now", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://needednowlt.substack.com/p/the-global-conversation-on-ai-in" } ], "stages": [ "higher" ], "summary": { "en": "Lodge reflects on learning, governance and the limits of chatbot-era reform after Digital Learning Week.", "zh": "Lodge 根据参与 UNESCO 数字学习周的经历，反思聊天机器人时期的改革是否足够。" }, "title": { "en": "The global conversation on AI in education is shifting. Where does it leave Australian higher education?", "zh": "AI 教育讨论转向学习与治理：Lodge 的一线观察" }, "topics": [ "learning", "assessment", "policy" ], "type": "opinion" }, { "body": { "en": [ "Selwyn calls for precise, contextual discussion of AI and warns against treating a power user’s experience as universal expertise. Constructive critique should ask what alternative forms of AI and educational practice people want." ], "zh": [ "Selwyn 提醒 AI 讨论者明确具体技术及其使用情境，不要把所有数字化问题都归于 AI，也不要用个人成功经验替别人决定怎样使用。他主张批判应结合历史和更多群体的声音，最终追问可行替代方案。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "selwyn-critical-ai-talk-20261002", "limits": { "en": "Opinionated critical commentary; it neither measures prevalence nor establishes that all AI uses share the same harms.", "zh": "明确带立场的批判性评论；不测量使用普遍程度，也不证明所有 AI 使用具有同样的危害。" }, "publishedAt": "2026-10-02", "sources": [ { "checkedAt": "2026-10-07T07:32:42.512802+00:00", "name": "Neil Selwyn · Critical Studies of Education & Technology", "scope": { "en": "Full nine-point public essay read; linked background works not independently reviewed.", "zh": "已读完整公开九点文章；所链接背景作品未独立审阅。" }, "status": "verified", "url": "https://criticaledtech.com/2026/10/02/things-to-bear-in-mind-when-talking-about-ai-some-notes-for-the-critically-minded/" } ], "stages": [ "school", "higher" ], "summary": { "en": "Selwyn calls for precise, contextual discussion of AI and warns against treating a power user’s experience as universal expertise. Constructive critique should ask what alternative forms of AI and educational practice people want.", "zh": "Selwyn 提醒 AI 讨论者明确具体技术及其使用情境，不要把所有数字化问题都归于 AI，也不要用个人成功经验替别人决定怎样使用。他主张批判应结合历史和更多群体的声音，最终追问可行替代方案。" }, "title": { "en": "Things to bear in mind when talking about AI", "zh": "谈 AI 教育，先说清是哪种技术、谁的经验" }, "topics": [ "theory", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Educator Zhang Shuling questions defining children’s worth through an ever-expanding list of abilities AI cannot yet perform. Useful capacities matter, but do not exhaust the purpose of education." ], "zh": [ "教育工作者张淑玲反思：当 AI 更会计算、创作与推理，教育是否只能不断为孩子增加“机器暂时做不到”的能力清单？她借“君子不器”提醒，人的价值不宜只按功能和竞争力衡量。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "zhang-human-value-ai-parenting-20261001", "limits": { "en": "The author’s educational and philosophical argument, not empirical evidence that skills training is harmful. Published in a Taiwanese parenting outlet; English title is an editorial translation.", "zh": "作者的教育哲学论述，不是证明技能训练有害的实证研究。文章发表于台湾教育媒体，英文标题为编辑翻译。" }, "publishedAt": "2026-10-01", "sources": [ { "checkedAt": "2026-10-07T07:32:42.513437+00:00", "name": "張淑玲 · 親子天下", "scope": { "en": "Full public article read through its final argument and author biography; publication and update dates both show 1 October 2026.", "zh": "已读公开文章全文至结尾和作者介绍；网页发布与更新日期均为2026年10月1日。" }, "status": "verified", "url": "https://www.parenting.com.tw/article/6002663" } ], "stages": [ "family", "school" ], "summary": { "en": "Educator Zhang Shuling questions defining children’s worth through an ever-expanding list of abilities AI cannot yet perform. Useful capacities matter, but do not exhaust the purpose of education.", "zh": "教育工作者张淑玲反思：当 AI 更会计算、创作与推理，教育是否只能不断为孩子增加“机器暂时做不到”的能力清单？她借“君子不器”提醒，人的价值不宜只按功能和竞争力衡量。" }, "title": { "en": "Must Children Keep Acquiring More Abilities in the AI Era? (translated title)", "zh": "孩子的价值，要靠比 AI 多一种能力来证明吗？" }, "topics": [ "theory", "learning" ], "type": "opinion" }, { "body": { "en": [ "Meyer highlights observations about teacher roles, curriculum fit, actionable suggestions and multilingual access, alongside disengaging uses. He asks how people actually use AI before treating capabilities as educational outcomes." ], "zh": [ "Meyer 从课堂走访提炼观察：明确教师角色、贴合课程、给出可调整的建议，以及支持多语言表达。他也强调，有些 AI 活动让学生只是机械完成任务。重点是把“产品能做什么”转为“师生实际怎样使用”。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "meyer-ai-learning-tour-20260930", "limits": { "en": "An author-selected interpretation of qualitative observations, not causal evidence. Meyer discloses involvement in Amplify’s AI feature; findings cannot represent every classroom or product.", "zh": "作者对定性观察的选择性解读，不是因果效果证据。Meyer 披露参与 Amplify 的 AI 功能开发；不能代表所有课堂或产品。" }, "publishedAt": "2026-09-30", "sources": [ { "checkedAt": "2026-10-07T07:32:42.546369+00:00", "name": "Dan Meyer · Mathworlds", "scope": { "en": "Full essay read; original project page and described interview/observation method checked. Full white paper and all product profiles not reviewed.", "zh": "已读完整公开文章并核对原项目介绍与访谈/观察方法；未读完整白皮书和全部产品介绍。" }, "status": "verified", "url": "https://danmeyer.substack.com/p/six-important-findings-from-the-ai" } ], "stages": [ "school" ], "summary": { "en": "Meyer highlights observations about teacher roles, curriculum fit, actionable suggestions and multilingual access, alongside disengaging uses. He asks how people actually use AI before treating capabilities as educational outcomes.", "zh": "Meyer 从课堂走访提炼观察：明确教师角色、贴合课程、给出可调整的建议，以及支持多语言表达。他也强调，有些 AI 活动让学生只是机械完成任务。重点是把“产品能做什么”转为“师生实际怎样使用”。" }, "title": { "en": "Six Important Findings from the AI in Action Learning Tour", "zh": "先看课堂怎样使用 AI，再问它是否有效" }, "topics": [ "teaching", "assessment", "tools" ], "type": "opinion" }, { "body": { "en": [ "The UNESCO IITE–Shanghai Open University project compares 36 teacher AI competency frameworks, examining competencies, implementation and assessment across policy, curriculum and teaching practice." ], "zh": [ "UNESCO IITE 与上海开放大学合作项目发布报告，比较36种国家、机构及学术教师 AI 能力框架，关注能力内容、实施场景与评价方法，并从政策、课程和教学层面讨论落地条件。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "unesco-iite-teacher-frameworks-20260928", "limits": { "en": "Based on the institutional announcement, not a full reading of the report. Framework comparisons do not demonstrate training effects on teachers or students.", "zh": "依据机构发布说明，未完整核读报告；框架比较不证明某种培训能改善教师表现或学生学习。" }, "publishedAt": "2026-09-28", "sources": [ { "checkedAt": "2026-10-07T07:32:42.620943+00:00", "name": "UNESCO IITE", "scope": { "en": "Official announcement and topic index read. The 28 September publication date comes from body text and index; a separately displayed 2 October date is not treated as first publication.", "zh": "官方新闻全文和主题索引已读；以正文和索引的9月28日为首次发布日期，不采用另出现的10月2日。" }, "status": "verified", "url": "https://iite.unesco.org/news/unesco-iite-has-released-the-analytical-report-teacher-ai-competency-frameworks/" } ], "stages": [ "school", "higher" ], "summary": { "en": "The UNESCO IITE–Shanghai Open University project compares 36 teacher AI competency frameworks, examining competencies, implementation and assessment across policy, curriculum and teaching practice.", "zh": "UNESCO IITE 与上海开放大学合作项目发布报告，比较36种国家、机构及学术教师 AI 能力框架，关注能力内容、实施场景与评价方法，并从政策、课程和教学层面讨论落地条件。" }, "title": { "en": "Teacher AI Competency Frameworks: UNESCO IITE’s comparative report", "zh": "UNESCO IITE 比较36种教师 AI 能力框架" }, "topics": [ "teaching", "assessment", "policy" ], "type": "research" }, { "body": { "en": [ "Jisc proposes noticing unproductive interactions, judging what AI can reliably handle, and redesigning workflows. Institutions should allocate learning time; a useful outcome may be recognizing when not to use AI." ], "zh": [ "Jisc 建议用“发现、判断、重设计”处理 AI 使用中的效率下降：识别反复提示和核查是否比手工更费时，判断工具可靠的范围，再调整整段工作流程。重要收获也可能是更清楚何时不该用 AI。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "jisc-ai-productivity-dip-20260922", "limits": { "en": "Practice advice and case reflection, not causal proof of a universal productivity curve. Continued AI use is not guaranteed to save time.", "zh": "实践建议与个案反思，不是证明普遍生产率曲线的因果研究；不能承诺坚持使用 AI 一定节时。" }, "publishedAt": "2026-09-22", "sources": [ { "checkedAt": "2026-10-07T07:32:42.621489+00:00", "name": "Jisc Artificial Intelligence team", "scope": { "en": "Official blog article read. Some Jisc main-site URLs returned 502 during verification; this blog entry was readable. A local link recheck on 7 October 2026 returned HTTP 403; earlier research-reader access succeeded. Access differs by route.", "zh": "官方博客正文已读。Jisc主站部分URL在本次核验中返回502，本文博客入口可读。 本地链接复测（2026-10-07）返回HTTP 403；此前研究读取可读，不代表所有访问路径都畅通。" }, "status": "limited", "url": "https://nationalcentreforai.jiscinvolve.org/wp/2026/09/22/notice-judge-redesign-working-through-the-ai-productivity-dip/" } ], "stages": [ "higher" ], "summary": { "en": "Jisc proposes noticing unproductive interactions, judging what AI can reliably handle, and redesigning workflows. Institutions should allocate learning time; a useful outcome may be recognizing when not to use AI.", "zh": "Jisc 建议用“发现、判断、重设计”处理 AI 使用中的效率下降：识别反复提示和核查是否比手工更费时，判断工具可靠的范围，再调整整段工作流程。重要收获也可能是更清楚何时不该用 AI。" }, "title": { "en": "Notice, Judge, Redesign: Working Through the AI Productivity Dip", "zh": "Jisc：AI 越用越费时，如何判断和调整工作流程？" }, "topics": [ "teaching", "tools" ], "type": "practice" }, { "body": { "en": [ "Barnett describes three competency areas: responsible use, workflow supervision, and leadership. Faculty, alumni and employers inform the proposed curriculum; direct engagement with communities remains important. Implementation was still incomplete.", "A useful planning question is where foundational competencies belong across a degree, and which require specialist depth. A discussion framework is not yet a completed reform." ], "zh": [ "Barnett 将公共卫生 AI 能力分成负责任使用、监督工作流程和领导治理三个层面。她介绍与教师、校友和雇主的讨论，并强调社区倾听仍需亲身练习。方案当时尚未全面落实。", "值得借鉴的是跨课程分工：哪些基础能力人人都要学，哪些留给专业方向？讨论框架与已经完成的课程改革，应分别判断。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "bruff-barnett-curriculum-20260922", "limits": { "en": "Interview self-report; no post-implementation outcomes.", "zh": "访谈自述；未报告实施后的学习成效。" }, "publishedAt": "2026-09-22", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Derek Bruff · Intentional Teaching", "scope": { "en": "Read curriculum, consultation and next-step transcript sections; audio unchecked.", "zh": "已读公开转录的课程框架、协商与后续计划；未核听音频。" }, "status": "verified", "url": "https://intentionalteaching.buzzsprout.com/2069949/episodes/19843873-curricular-responses-to-ai-with-deborah-barnett" } ], "stages": [ "higher" ], "summary": { "en": "Coordinating AI competencies across a professional program.", "zh": "公共卫生培养方案如何协调 AI 能力与专业价值？" }, "title": { "en": "Curricular Responses to AI with Deborah Barnett", "zh": "把 AI 纳入专业培养方案，而不只写进单门课政策" }, "topics": [ "teaching", "policy" ], "type": "practice" }, { "body": { "en": [ "Liu reports on a program at Kaohsiung Girls’ Senior High School in Taiwan emphasizing problem selection, collaboration, AI-supported making and explaining concepts to other learners." ], "zh": [ "报道介绍台湾高雄女中的 AI 电资教育：学生从实际需求中选择项目，再学习描述需求、整合想法、合作制作，也通过为其他学生设计课程检验理解。教师强调理解原理与保有思考。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "liu-ai-school-problem-finding-20260920", "limits": { "en": "A single-school report and interview views, without controlled or long-term evaluation. Not a general effect or mainland China policy; a pictured competition award is explicitly dated 2025.", "zh": "单校报道与访谈观点，无对照组或长期效果测量。不能外推为普遍效果或大陆政策；文章配图中的竞赛奖项明确为2025年。" }, "publishedAt": "2026-09-20", "sources": [ { "checkedAt": "2026-10-07T07:32:42.819123+00:00", "name": "劉怡彣 · 國立教育廣播電台／獨立評論", "scope": { "en": "Full article read. Linked radio program not listened to; this is article-based, not an audio transcript. English title is an editorial translation.", "zh": "已读完整发表文章。未收听链接的广播节目，本条基于文章而非音频转录；英文标题为编辑翻译。" }, "status": "verified", "url": "https://opinion.cw.com.tw/blog/profile/480/article/17780" } ], "stages": [ "school" ], "summary": { "en": "Liu reports on a program at Kaohsiung Girls’ Senior High School in Taiwan emphasizing problem selection, collaboration, AI-supported making and explaining concepts to other learners.", "zh": "报道介绍台湾高雄女中的 AI 电资教育：学生从实际需求中选择项目，再学习描述需求、整合想法、合作制作，也通过为其他学生设计课程检验理解。教师强调理解原理与保有思考。" }, "title": { "en": "What Should We Teach Children When AI Can Do So Much? (translated title)", "zh": "AI 项目课的重点：发现问题、合作与教会别人" }, "topics": [ "teaching", "learning", "tools" ], "type": "practice" }, { "body": { "en": [ "Mollick argues that expertise, breadth, taste and agency help people use AI capabilities. His experiments invite a question: how do learners acquire the knowledge and judgment needed to direct and evaluate increasingly capable tools?" ], "zh": [ "Mollick 用个人 AI 制作实验讨论能力与使用之间的落差，主张深度知识、广泛知识、品味与主动探索仍能决定合作质量。教育上的一个阅读问题是：学习者如何获得指导和评价工具所需的知识与判断？" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "mollick-the-overhang-20260918", "limits": { "en": "Author argument illustrated by personal demonstrations; no controlled learning-outcome evidence. The educational question is editorial interpretation, not a tested curriculum.", "zh": "作者论点与个人示范，不是受控学习效果试验。教育启发是编辑提出的阅读问题，不是作者验证过的课程。" }, "publishedAt": "2026-09-18", "sources": [ { "checkedAt": "2026-10-07T07:32:43.061908+00:00", "name": "Ethan Mollick · One Useful Thing", "scope": { "en": "Full public article read, including the four-advantages argument and conclusion. Linked demos and research not independently evaluated; videos not watched.", "zh": "已读完整公开文章，包括四种优势与结论；链接演示和引用研究未独立评价，视频未观看。" }, "status": "verified", "url": "https://www.oneusefulthing.org/p/the-overhang" } ], "stages": [ "higher", "general" ], "summary": { "en": "Mollick argues that expertise, breadth, taste and agency help people use AI capabilities. His experiments invite a question: how do learners acquire the knowledge and judgment needed to direct and evaluate increasingly capable tools?", "zh": "Mollick 用个人 AI 制作实验讨论能力与使用之间的落差，主张深度知识、广泛知识、品味与主动探索仍能决定合作质量。教育上的一个阅读问题是：学习者如何获得指导和评价工具所需的知识与判断？" }, "title": { "en": "The Overhang", "zh": "当 AI 能力超过实际用法，学习者还需要什么？" }, "topics": [ "theory", "learning", "tools" ], "type": "opinion" }, { "body": { "en": [ "Sumeracki explains Butler et al.’s (2017) experiments contrasting repeated and varied application during retrieval practice. Her account invites asking whether an AI activity makes students retrieve, explain and apply concepts across contexts." ], "zh": [ "Sumeracki 解读 Butler 等（2017）关于检索练习的实验，讨论在不同例子中回忆和应用概念与后续新情境表现的关系。用于 AI 学习设计时，可追问工具是否让学生真正检索、解释和迁移，而非不断提供相似答案。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "sumeracki-varied-retrieval-20260917", "limits": { "en": "A 2026 explainer of Butler et al. (2017), not a new AI experiment. The underlying paper was not read and quantitative findings were not independently validated. The AI connection is an editorial inference.", "zh": "这是2026年对 Butler 等（2017）的研究解读，不是新的 AI 实验。原始论文未读，数量结果未独立验证；AI 应用是编辑推论。" }, "publishedAt": "2026-09-17", "sources": [ { "checkedAt": "2026-10-07T07:32:43.200133+00:00", "name": "Megan Sumeracki · The Learning Scientists", "scope": { "en": "Full public explainer, including four experiments and reference, read. Underlying DOI retrieval was robots-blocked; no bypass attempted.", "zh": "已读完整公开解读及四项实验介绍与参考文献。原论文 DOI 访问受 robots.txt 阻止，未绕过限制。" }, "status": "verified", "url": "https://www.learningscientists.org/blog/2026/9/17" } ], "stages": [ "higher", "general" ], "summary": { "en": "Sumeracki explains Butler et al.’s (2017) experiments contrasting repeated and varied application during retrieval practice. Her account invites asking whether an AI activity makes students retrieve, explain and apply concepts across contexts.", "zh": "Sumeracki 解读 Butler 等（2017）关于检索练习的实验，讨论在不同例子中回忆和应用概念与后续新情境表现的关系。用于 AI 学习设计时，可追问工具是否让学生真正检索、解释和迁移，而非不断提供相似答案。" }, "title": { "en": "Varying Retrieval Practice to Promote Transfer of Learning", "zh": "练习不只要重复，也要换情境：迁移学习的启发" }, "topics": [ "theory", "learning", "assessment" ], "type": "news" }, { "body": { "en": [ "Demonstrations isolate one audible and visible change while hiding irrelevant setup. Chongqing writes the course outlines himself, then uses AI to help build web slides and demonstrations.", "His reading process returns from retrieved passages to original texts and explores connections across books. The account makes a useful distinction between finding snippets and understanding a sustained argument." ], "zh": [ "重轻把声音变化设计成一个可听、可见的动作，隐藏无关操作，让听众集中感知。课程提纲由他自己写，AI 再帮助制作演示与网页幻灯。", "阅读时，他从 AI 指向的段落回到原文，并主动跨书寻找关联。这提醒备课者：检索到片段和真正把握整本书的论证，应当区分。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "shishufeng-course-ai-reading-20260917", "limits": { "en": "A personal preparation workflow; capability claims are bounded to the described system and experience, not all AI.", "zh": "个人备课工作流；对 AI 阅读能力的概括限于其所述系统与经历，不能扩展为所有模型的固定边界。" }, "publishedAt": "2026-09-17", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "重轻、咪仔 · 诗梳风 EP29", "scope": { "en": "Recovered transcript; close-read demonstration design, outlines, cross-book reading and retrieval limitations. Tools/course not independently tested. The publisher identifies this as an audio-derived transcript and warns of typographical/recognition errors; audio was not checked.", "zh": "恢复官方逐字稿，细读教具设计、提纲流程、跨书阅读与检索局限段落；未运行工具或看原课。 官方说明逐字稿整理自原始音频，可能有错别字或识别误差；本轮未核听音频。" }, "status": "limited", "url": "https://shishufeng.com/episode/20th_century_instrument_2/transcript" } ], "stages": [ "general", "higher" ], "summary": { "en": "The second retrospective describes AI-built demonstrations and source navigation, while exposing the limits of keyword retrieval.", "zh": "课程复盘下集展示 AI 教具与资料阅读流程，也坦陈关键词检索难以把握跨段落的大论证。" }, "title": { "en": "Reading Across Fields and Working with AI: A Course Retrospective — Shishufeng EP29", "zh": "AI 能做教具，课程的解释仍要自己负责" }, "topics": [ "teaching", "learning", "tools" ], "type": "practice" }, { "body": { "en": [ "One question concerns automated judgments of competence; another concerns evidence of student capability under uncontrolled AI access.", "Dawson argues that wearable systems expose the fragility of designs relying on clean separation from AI." ], "zh": [ "前一个问题追问机器评分是否可信；后一个问题追问提交或现场表现还能证明什么。", "他认为可穿戴 AI 正在削弱“把学生与 AI 分开”这一前提，邀请医学教育的效度研究参与讨论。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "dawson-two-validities-20260917", "limits": { "en": "Conceptual commentary with no new dataset; device and market claims were not independently checked.", "zh": "概念评论，无新数据；设备能力与市场数字未独立核实。" }, "publishedAt": "2026-09-17", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Phillip Dawson · Medical Education", "scope": { "en": "Read the open-access commentary, declarations and publication metadata.", "zh": "阅读开放获取评论全文、声明与元数据。" }, "status": "verified", "url": "https://asmepublications.onlinelibrary.wiley.com/doi/10.1111/medu.70320" } ], "stages": [ "higher" ], "summary": { "en": "Dawson separates validating AI assessors from validating assessment when students have AI access.", "zh": "Dawson 区分机器如何评学生，与学生可使用 AI 时如何判断其能力。" }, "title": { "en": "Two validity conversations in a time of AI", "zh": "AI 评价中的两种效度问题不能混为一谈" }, "topics": [ "assessment", "theory" ], "type": "opinion" }, { "body": { "en": [ "Wexler opposes indiscriminate rejection of educational technology, separating teacher support from tools that replace pupils’ thinking. Judy and Amira illustrate questions of review, practice opportunities and family communication.", "This is a skeptic’s conditional endorsement. Several efficacy and privacy statements come from company representatives and require separate verification. The article notes a 28 September update." ], "zh": [ "Wexler 反对无差别排除教育技术，认为教师备课支持与直接替学生完成思考的工具应分别判断。她以 Judy 和 Amira 为例，讨论审核能力、练习机会及家长对数据使用的疑问。", "其价值是展示一个怀疑者的有条件支持；产品功效和隐私说法中有相当部分来自企业负责人，不能替代独立评估或合同审查。文章于 9 月 28 日补充内容。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "wexler-selective-edtech-20260913", "limits": { "en": "Author has Writing Revolution ties; company privacy and screening claims not verified. No diagnostic advice.", "zh": "作者与 The Writing Revolution 有关联；未验证企业隐私和识别准确性主张，不作诊断建议。" }, "publishedAt": "2026-09-13", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Natalie Wexler / Minding the Gap", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://nataliewexler.substack.com/p/dont-throw-ed-tech-babies-out-with" } ], "stages": [ "school", "family" ], "summary": { "en": "Teacher workload, student thinking, product evidence and family communication are treated separately.", "zh": "作者把教师减负、学生思考、产品证据和家校沟通拆开讨论。" }, "title": { "en": "Don’t Throw Ed Tech Babies Out With the Bathwater", "zh": "谨慎采用 AI，也要区分具体用途" }, "topics": [ "teaching", "learning", "policy", "tools" ], "type": "opinion" }, { "body": { "en": [ "One session contracted to a tiny part of its original scope while gaining a stronger explanatory focus. Preparation involved deciding what to discard, not merely adding material.", "Mizai’s vivid memory of live performance reveals what a participant actually retained. The exchange supports discussion of coverage, understanding and the value of perceptible experience." ], "zh": [ "重轻回忆，有一讲最终只保留原计划极小的一部分，却让核心解释更有穿透力。备课中的困难包含发现什么值得舍弃，而不只是继续增加材料。", "咪仔对现场演奏记忆深刻，也让讲述者看见听众实际带走了什么。这个案例适合讨论内容覆盖、理解与可感知经验之间的关系。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "shishufeng-course-purpose-20260911", "limits": { "en": "One creator/participant retrospective. Vivid recall does not establish deep learning; English heading is editorial translation.", "zh": "单门公开课程的创作者与学员复盘；记忆鲜明不能自动等同深度学习。英文题为编辑译题。" }, "publishedAt": "2026-09-11", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "重轻、咪仔 · 诗梳风 EP28", "scope": { "en": "Recovered transcript; close-read course framing, redesign of the electronic-instrument session and participant feedback. Course and audio not watched/listened. The publisher identifies this as an audio-derived transcript and warns of typographical/recognition errors; audio was not checked.", "zh": "恢复官方逐字稿，细读课程动机、电子乐器一讲的重构与现场演示反馈；未听完整音频或原课程。 官方说明逐字稿整理自原始音频，可能有错别字或识别误差；本轮未核听音频。" }, "status": "limited", "url": "https://shishufeng.com/episode/20th_century_instrument_1/transcript" } ], "stages": [ "general", "higher" ], "summary": { "en": "Chongqing and participant Mizai revisit how a music course found its themes and used live demonstrations.", "zh": "重轻与听课者咪仔复盘“二十世纪的乐器”，讨论课程主题如何在准备中改变，以及现场演示的价值。" }, "title": { "en": "Motivation, Themes and Practice: A Course Retrospective — Shishufeng EP28", "zh": "一门音乐课怎样找到值得讲的核心？" }, "topics": [ "teaching", "learning" ], "type": "practice" }, { "body": { "en": [ "Across 200 higher education institutions in 19 countries, 87% reported AI use in at least one area, while 26% had a formal AI strategy. The report calls for stronger governance and institutional capability." ], "zh": [ "UNESCO IESALC 与 UNU-IAS 对19国200所高校的研究中，87%的受访机构报告在至少一个领域使用 AI，26%拥有正式 AI 战略。官方介绍建议将采用与治理、数字素养及组织能力建设衔接。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "unesco-iesalc-lac-ai-20260908", "limits": { "en": "Institution-reported adoption, not causal evidence of improved learning. Sampling methods were not fully checked; percentages are not estimates for all universities or regions.", "zh": "机构报告的采用情况，不是改善学习的因果证据。未完整核验抽样方法，不将比例当作所有地区高校的总体估计。" }, "publishedAt": "2026-09-08", "sources": [ { "checkedAt": "2026-10-07T07:32:43.628034+00:00", "name": "UNESCO IESALC / UNU-IAS", "scope": { "en": "Official report description read in full. Figures come from the publisher summary; the full report was not reviewed.", "zh": "官方报告介绍页全文已读；数字据出版机构简介，未完整核读报告。" }, "status": "verified", "url": "https://www.iesalc.unesco.org/en/articles/ai-implementation-higher-education-latin-america-and-caribbean" } ], "stages": [ "higher" ], "summary": { "en": "Across 200 higher education institutions in 19 countries, 87% reported AI use in at least one area, while 26% had a formal AI strategy. The report calls for stronger governance and institutional capability.", "zh": "UNESCO IESALC 与 UNU-IAS 对19国200所高校的研究中，87%的受访机构报告在至少一个领域使用 AI，26%拥有正式 AI 战略。官方介绍建议将采用与治理、数字素养及组织能力建设衔接。" }, "title": { "en": "AI Implementation in Higher Education in Latin America and the Caribbean", "zh": "拉美与加勒比高校 AI 调查：采用速度快于制度建设" }, "topics": [ "policy", "teaching" ], "type": "research" }, { "body": { "en": [ "This UNESCO collection brings together 26 think pieces on agency, young people’s development, pedagogy, expression, public governance and educational futures. It broadens discussion beyond tool capabilities." ], "zh": [ "UNESCO 汇集26篇思想短文，把讨论扩展到人的自主性、青少年发展、教学评价、表达与质疑、公共治理及教育未来想象。它提供讨论议题，帮助把工具能力与教育目的放在一起思考。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "unesco-algorithm-room-20260908", "limits": { "en": "A collection of think pieces, not a systematic review or effectiveness trial. The essays are not treated as one consensus.", "zh": "思想短文集，不是系统综述或效果试验；未将26篇文章概括成单一共识。" }, "publishedAt": "2026-09-08", "sources": [ { "checkedAt": "2026-10-07T07:32:43.281295+00:00", "name": "UNESCO", "scope": { "en": "Official publication description read; the 26 essays were not individually reviewed. Original publication: 8 September; update: 9 September.", "zh": "已读官方出版介绍；未逐篇核读整部文集。原发布为9月8日，不将9月9日更新当首次发布。" }, "status": "verified", "url": "https://www.unesco.org/en/articles/algorithm-room-seeing-and-confronting-implications-ai-future-education" } ], "stages": [ "school", "higher" ], "summary": { "en": "This UNESCO collection brings together 26 think pieces on agency, young people’s development, pedagogy, expression, public governance and educational futures. It broadens discussion beyond tool capabilities.", "zh": "UNESCO 汇集26篇思想短文，把讨论扩展到人的自主性、青少年发展、教学评价、表达与质疑、公共治理及教育未来想象。它提供讨论议题，帮助把工具能力与教育目的放在一起思考。" }, "title": { "en": "The algorithm in the room: AI and the future of education", "zh": "UNESCO《房间里的算法》：教育的六个未决问题" }, "topics": [ "theory", "assessment", "policy" ], "type": "opinion" }, { "body": { "en": [ "Christodoulou rejects generic exercises as a route to universally transferable creativity or critical thinking. She contrasts transitions between knowledge-adjacent jobs with superficially similar tasks in unrelated domains.", "For AI education, this challenges replacing subject knowledge with general judgment skills. It belongs beside competing positions, not under a headline claiming a new experimental discovery." ], "zh": [ "Christodoulou 否认存在能普遍训练的通用创造力、沟通和批判思考能力。她用职业转换区分知识相近时的迁移与跨领域的表面相似，并追溯这种观念的历史。", "对于 AI 课程，真正的分歧是：教“通用判断力”能否替代扎实的学科积累？应与更支持跨情境能力培养的作者并读，不把这个强结论包装成新的实验证明。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-transferable-skills-20260906", "limits": { "en": "Strong theoretical position, not a new experiment; its referenced evidence review was not fully audited.", "zh": "作者的强理论立场；本文不是新试验，也未逐篇复核其证据综述。" }, "publishedAt": "2026-09-06", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/what-if-transferable-skills-dont" } ], "stages": [ "school", "higher", "general" ], "summary": { "en": "A deliberately strong argument locates reasoning and problem-solving in domain knowledge.", "zh": "一篇强立场文章要求把批判思考与解决问题落实到具体领域知识。" }, "title": { "en": "What if transferable skills don’t exist?", "zh": "“可迁移技能”究竟迁移了什么？" }, "topics": [ "theory", "learning", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Wexler argues that appropriate AI use depends on what pupils are supposed to practice. Outlining, for example, can be the learning task rather than disposable overhead.", "Discussing teacher-preparation frameworks and an interview with Deans for Impact, she calls for stronger instructional judgment, materials and student understanding of effort. A planned framework is not an existing policy." ], "zh": [ "Wexler 认为，判断 AI 是否有益需要先明确学生应该练习什么。她尤其反对把提纲等任务轻率称为低价值负担，因为组织论证本身可能就是学习目标。", "她讨论教师培养框架，并采访 Deans for Impact 负责人，强调培养教师判断、改善教材和向学生说明认知努力的价值。规划中的框架不应被当作已发布政策。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "wexler-ai-cognitive-science-20260902", "limits": { "en": "Read original commentary and interview reporting; not every linked survey, policy or study was independently verified.", "zh": "已读原始评论与采访内容；外链调查、政策及研究未全部独立复核。" }, "publishedAt": "2026-09-02", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Natalie Wexler / Minding the Gap", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://nataliewexler.substack.com/p/how-should-ai-be-used-in-schools" } ], "stages": [ "school", "higher" ], "summary": { "en": "Examples of planning and writing ask which cognitive tasks can appropriately be delegated.", "zh": "文章以备课、提纲与写作为例，追问究竟哪些认知工作可以外包。" }, "title": { "en": "How Should AI Be Used in Schools?", "zh": "教师 AI 素养需要理解学习，而不只是会操作" }, "topics": [ "teaching", "theory", "policy" ], "type": "opinion" }, { "body": { "en": [ "Warner supports ethics teaching but doubts that moral identity alone can overcome competitive incentives, time pressure and inconsistent AI rules. He argues for meaningful practice and intrinsic motivation rather than concentrating institutional effort on detection.", "The discussion question is whether course incentives reinforce the honest choices educators demand. Structural explanation need not erase individual responsibility." ], "zh": [ "Warner 支持伦理教育，却质疑单靠道德身份就能抵消成绩竞争、时间压力和不一致的 AI 规则。他主张改变学习体验，让学生投入有意义的练习，而非把全部精力放在阻止作弊。", "值得讨论的是责任如何分配：要求学生作出诚实选择的同时，课程是否奖励了相反行为？这是制度论证，不是免除个人责任。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "warner-morality-cheating-20260828", "limits": { "en": "Opinion; cited studies and student testimony not independently checked.", "zh": "评论；引述研究和学生说法未在本轮独立核验。" }, "publishedAt": "2026-08-28", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "John Warner · Inside Higher Ed / Just Visiting", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://www.insidehighered.com/opinion/columns/just-visiting/2026/08/28/integrity-great-you-cant-eat-good-morals" } ], "stages": [ "higher" ], "summary": { "en": "Reframing misconduct around incentives and learning.", "zh": "把问题放回竞争、动机和课程结构。" }, "title": { "en": "Moral Instruction Will Not Stop Cheating", "zh": "只加强道德教育，能减少 AI 代写吗？" }, "topics": [ "assessment", "teaching", "theory" ], "type": "opinion" }, { "body": { "en": [ "Bernards combines draft labs, peer responses, reflection, individual emails and student-designed tutoring prompts. She also assigns 80% of the course grade to proctored exams, acknowledging conflict with her assessment preferences.", "Read the support structure alongside its labor and assessment costs. Fast completion alone cannot establish misconduct, and online access involves more than examination convenience." ], "zh": [ "Bernards 描述草稿实验、同伴回应、反思、每周个别邮件及自建辅导机器人练习。与此同时，她把监考考试提高到总成绩的 80%，承认这违背自己的评价偏好。", "这份案例值得按支持成本与评价代价一起读。答题很快不能单独证明作弊；线上可及性也不应只靠最终考试是否方便来衡量。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "bruff-bernards-online-math-20260825", "limits": { "en": "Teacher account; speed is not proof of cheating.", "zh": "教师自述；速度推断不构成作弊证据。" }, "publishedAt": "2026-08-25", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Derek Bruff · Intentional Teaching", "scope": { "en": "Read substantive teaching transcript sections; audio unchecked.", "zh": "已读公开转录主要教学段落；未核听音频。" }, "status": "verified", "url": "https://intentionalteaching.buzzsprout.com/2069949/episodes/19696094-ai-aware-online-math-teaching-with-jessica-bernards" } ], "stages": [ "higher" ], "summary": { "en": "Rich learning support alongside high-stakes proctoring.", "zh": "互评、反思和个别联系，与监考压力并存。" }, "title": { "en": "AI-Aware Online Math Teaching with Jessica Bernards", "zh": "在线数学课：支持学习与高权重监考的拉扯" }, "topics": [ "teaching", "learning", "assessment" ], "type": "practice" }, { "body": { "en": [ "Furze proposes two contextual questions: is this a good use of the technology, and of the learner’s own thinking? He examines accessibility, outsourcing and assessment, questioning whether fixed categories capture individual needs." ], "zh": [ "Furze 建议同时追问技术是否适合任务、使用方式是否有助于自己的思考，把判断放回具体学生、任务与时刻。语音转写、构思或整理格式可能帮助表达，也可能绕过需要练习的能力。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "furze-good-use-brain-20260824", "limits": { "en": "Practitioner reflection and examples, not a validated decision tool or comparative classroom study.", "zh": "实践者反思与示例，不是经过验证的决策工具或课堂比较试验。" }, "publishedAt": "2026-08-24", "sources": [ { "checkedAt": "2026-10-07T07:32:43.933950+00:00", "name": "Leon Furze", "scope": { "en": "Full public article read through final reflective questions; embedded media not watched.", "zh": "已读完整公开文章至最后的反思问题；内嵌媒体未观看。" }, "status": "verified", "url": "https://leonfurze.com/2026/08/24/is-this-a-good-use-of-the-technology-is-it-a-good-use-of-your-brain/" } ], "stages": [ "school", "higher" ], "summary": { "en": "Furze proposes two contextual questions: is this a good use of the technology, and of the learner’s own thinking? He examines accessibility, outsourcing and assessment, questioning whether fixed categories capture individual needs.", "zh": "Furze 建议同时追问技术是否适合任务、使用方式是否有助于自己的思考，把判断放回具体学生、任务与时刻。语音转写、构思或整理格式可能帮助表达，也可能绕过需要练习的能力。" }, "title": { "en": "Is This a Good Use of the Technology? Is It a Good Use of Your Brain?", "zh": "AI 用得好，还要问：对这次学习有帮助吗？" }, "topics": [ "learning", "assessment", "theory" ], "type": "opinion" }, { "body": { "en": [ "Watkins distinguishes independent drafting, critical study of AI and advanced experimentation across his courses. He argues that institutions owe faculty time, training and support rather than merely delegating consequential choices.", "A department can ask whether each boundary follows a learning purpose, and who carries the work of making it real." ], "zh": [ "Watkins 对照自己的初级写作、数字媒体和高年级课程：保护独立起草、分析 AI，或探索高级应用，各有不同目的。他同时要求学校为教师提供时间、培训与稳定支持，而非只下放责任。", "可用于院系讨论：政策是否解释了某阶段为什么需要或不需要 AI？执行这些选择的劳动，又由谁承担？" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "watkins-duty-care-20260823", "limits": { "en": "Normative guidance, not legal interpretation or an outcome study.", "zh": "规范性建议与实践设想；不是法律解释或效果评估。" }, "publishedAt": "2026-08-23", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Marc Watkins · Rhetorica", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://marcwatkins.substack.com/p/our-duty-of-care-in-the-age-of-ai" } ], "stages": [ "higher" ], "summary": { "en": "Course purposes and faculty working conditions belong together.", "zh": "学生学习目标与教师工作条件必须一起考虑。" }, "title": { "en": "Our Duty of Care in the Age of AI", "zh": "不同课程，为什么需要不同的 AI 边界？" }, "topics": [ "teaching", "policy", "theory" ], "type": "opinion" }, { "body": { "en": [ "Willingham treats assignments as practice in mental processes rather than merely products. He proposes delegating tasks only after students can perform them well and the task is no longer an instructional target.", "He explicitly excludes purpose-built instructional software and teacher use. The unresolved practical question is how to establish mastery and decide which practice still matters." ], "zh": [ "Willingham 认为学生作业的核心通常是完成过程中的思考，而非成品。他提出，只有学生已经熟练、且不再属于当前学习目标的任务，才适合交给 AI。", "文中明确排除专门设计的教学软件和教师用 AI。争议在于怎样判断掌握程度及哪些练习仍有价值；一个好看的作品或一次自称“会了”并不能给出答案。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "willingham-master-before-delegating-20260820", "limits": { "en": "A proposed principle, not policy or a validated mastery threshold; district descriptions are not treated as current rules.", "zh": "作者建议，不是统一政策或已验证的掌握阈值；不转述其学区描述为现行规则。" }, "publishedAt": "2026-08-20", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daniel Willingham", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://dtwuva.substack.com/p/students-should-only-use-ai-for-things" } ], "stages": [ "school", "higher", "family" ], "summary": { "en": "The proposed restriction concerns independent general-purpose AI use, not all instructional software.", "zh": "限制针对学生独立使用通用 AI，不能扩大成反对一切教学软件。" }, "title": { "en": "Students should only use AI for things they already know how to do well.", "zh": "先熟练掌握，再考虑把任务交给 AI" }, "topics": [ "learning", "theory", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Using public descriptions, Wexler sees promise in AlphaWrite’s structured practice and feedback but questions separating writing topics from curriculum content. She explicitly had neither accessed the app nor visited a school.", "The dispute treats writing as both skill and a means of learning subject matter. Reported test-growth multiples should not be equated with independent writing or broad transfer." ], "zh": [ "Wexler 根据公开介绍分析 AlphaWrite 的句子、段落与提纲训练，认可练习和即时反馈的潜力，却质疑写作内容与课程分离。她明确说明尚未试用软件或到访学校。", "争论核心是写作既是技能，也能帮助理解学科内容。学校公布的成长倍数不能直接等同于独立写作能力或广泛迁移；本文的产品描述仍依赖他人公开说明。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "wexler-alphawrite-curriculum-20260817", "limits": { "en": "Not a hands-on product test or independent school evaluation; author has ties to The Writing Revolution.", "zh": "非产品实测或学校独立评估；作者与 The Writing Revolution 有关联。" }, "publishedAt": "2026-08-17", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Natalie Wexler / Minding the Gap", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://nataliewexler.substack.com/p/how-alpha-school-teaches-writing" } ], "stages": [ "school", "family" ], "summary": { "en": "The critique asks whether interest-driven writing loses opportunities to consolidate curriculum knowledge.", "zh": "作者关注按兴趣选题是否削弱写作对课程知识的促进。" }, "title": { "en": "How Alpha School Teaches Writing", "zh": "AlphaWrite 争论：写作技巧应与哪些内容连接？" }, "topics": [ "teaching", "learning", "theory" ], "type": "opinion" }, { "body": { "en": [ "A Yuan-history example moves from AI-assisted disciplinary mapping to multilingual keywords and human selection of books, papers and primary records. Searching itself becomes part of learning.", "Liushou challenges excessive preparation and explains how new questions arise during writing. The exchange helps distinguish assembling materials from developing judgment about what matters." ], "zh": [ "节目以元代研究为例，先借 AI 勾勒学科问题，再生成多语言关键词，最后自己寻找和判断书籍、论文与史料。检索被视为学习本身的一部分。", "六兽追问资料积累的边界，指出创作途中也会生出新问题。教师可借此讨论研究问题如何变化，以及资料数量为何不能替代取舍。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "shishufeng-research-literacy-20260813", "limits": { "en": "Creator practice, not validated retrieval coverage. AI syllabi and references need checking; source prompts were not executed.", "zh": "创作者方法，未验证检索覆盖率；AI 课纲与引用仍需核查。没有执行节目中的提示词或下载建议。" }, "publishedAt": "2026-08-13", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "汉洋、六兽 · 诗梳风 EP27", "scope": { "en": "Recovered transcript; close-read syllabus/keyword workflow, source selection and over-preparation discussion. No audio or replication. The publisher identifies this as an audio-derived transcript and warns of typographical/recognition errors; audio was not checked.", "zh": "恢复官方逐字稿，细读课纲与关键词流程、史料取舍和过度准备讨论；未听音频或验证演示结果。 官方说明逐字稿整理自原始音频，可能有错别字或识别误差；本轮未核听音频。" }, "status": "limited", "url": "https://shishufeng.com/episode/zhaoziliao/transcript" } ], "stages": [ "higher", "general" ], "summary": { "en": "Hanyang and Liushou discuss moving from an unfamiliar topic to a syllabus, search terms and original sources.", "zh": "汉洋与六兽讨论如何从一个陌生主题走向课纲、检索词与原始资料，也讨论资料过多怎样妨碍创作。" }, "title": { "en": "Finding Sources in the AI Era — Shishufeng EP27", "zh": "找资料，也是在建立对学科的理解" }, "topics": [ "learning", "teaching", "tools" ], "type": "practice" }, { "body": { "en": [ "Gulya asks students to design projects, predict processes, choose AI involvement and create criteria. Brief comment exchanges reduce resubmission burden; an optional chatbot scaffolds rubric creation. He remains uncertain how to assess metacognition reliably.", "The transferable question is how to preserve learner choice while checking common goals and keeping feedback manageable." ], "zh": [ "Gulya 让学生提出项目、预测过程、决定 AI 的位置，并建立自己的评价标准。教师用简短往返讨论代替反复交整份作业；可选机器人帮助整理标准。他也承认尚未解决如何可靠评价元认知。", "可迁移的问题是：给学生自主权之后，教师怎样确认目标覆盖并控制反馈负担？这比只允许或禁止某一种工具更具体。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "bruff-gulya-student-agency-20260813", "limits": { "en": "Classroom account; learning assurance and workload gains unverified.", "zh": "课堂经验；未独立验证学习保障或工作量下降。" }, "publishedAt": "2026-08-13", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Derek Bruff · Intentional Teaching", "scope": { "en": "Read substantial practice and unresolved-problem transcript sections.", "zh": "已读公开转录的主体实践与未解决问题。" }, "status": "verified", "url": "https://intentionalteaching.buzzsprout.com/2069949/episodes/19640762-ai-and-student-agency-with-jason-gulya" } ], "stages": [ "higher" ], "summary": { "en": "Student choice within shared course goals.", "zh": "把 AI 选择权放进仍有共同目标的课程。" }, "title": { "en": "AI and Student Agency with Jason Gulya", "zh": "学生自己设计项目、过程与评价标准" }, "topics": [ "teaching", "assessment", "learning" ], "type": "practice" }, { "body": { "en": [ "He proposes making evidence checking and reasoning revisable components rather than rewarding only polished final products.", "The essay connects grading incentives with trust and faculty support. Its broad claims favouring alternative grading remain the author’s argument." ], "zh": [ "他建议把核查证据、解释推理等过程分成可以获得反馈并再次尝试的环节。", "文章将评分制度与师生信任联系起来，是对高教改革的立场论证，不是对替代评分所有效果的验证。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "talbert-horizon-grading-20260810", "limits": { "en": "Opinion and experiential inference; cited studies were not individually audited.", "zh": "评论与经验推断；文中引用研究未在此逐项复核。" }, "publishedAt": "2026-08-10", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Robert Talbert · Grading for Growth", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://gradingforgrowth.com/p/three-ways-alternative-grading-addresses" } ], "stages": [ "higher" ], "summary": { "en": "Talbert argues that changing tasks without changing grading leaves a core incentive problem untouched.", "zh": "Talbert 从教学支持管理者视角，主张把反馈与再试机会纳入改革。" }, "title": { "en": "Three ways alternative grading addresses the EDUCAUSE Horizon report", "zh": "为何 AI 时代不仅要改任务，也要改评分？" }, "topics": [ "assessment", "policy", "theory" ], "type": "opinion" }, { "body": { "en": [ "Mills worries that generated materials may feel alienating and proposes explaining human effort and reasons alongside AI disclosures. She explicitly calls for research on students’ responses to different disclosure approaches.", "A useful materials audit asks whether students can identify responsibility and educational intent, not just a list of technologies." ], "zh": [ "Mills 担心大量生成的教材可能让学生感到被忽视。她主张除了说明用了哪些 AI，还解释哪些工作坚持亲自完成、为什么这样选择，并承认需要研究不同披露方式如何影响学生。", "这可以转化为一次课程材料审查：学生看得到谁对内容负责吗？披露是否表达了教育意图，而不只是列技术清单？" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "mills-human-intent-materials-20260807", "limits": { "en": "Argument and talk reflection; no demonstrated learning benefit.", "zh": "观点与讲座反思；未证明披露会提升学习。" }, "publishedAt": "2026-08-07", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Anna Mills · Substack", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://annamills.substack.com/p/lets-emphasize-human-effort-and-intent" } ], "stages": [ "higher" ], "summary": { "en": "Disclosure as a question of educational relationships.", "zh": "透明度也关乎学生是否感受到教师在场。" }, "title": { "en": "Let’s emphasize human effort and intent in educational materials", "zh": "教材使用 AI，披露还需要解释人的投入" }, "topics": [ "teaching", "policy", "theory" ], "type": "opinion" }, { "body": { "en": [ "Students develop projects through repeated critique. A one-day design exercise starts with individual analog ideation, moves through discussion and planning, and only then introduces rapid prototyping.", "Zarrow argues that building too early hardens weak concepts. The account offers concrete ways to teach discernment and include safety and ethics in assessment, while leaving the durability of these gains open." ], "zh": [ "课程把学生作品交给同伴与教师讨论，逐步澄清创作目标与质量标准。限时设计先让个人用纸笔发散，再组队规划，最后才用 AI 制作原型。", "作者观察到，过早写代码会让学生依恋尚未成熟的想法。对教学的启发是，把判断的练习放进修改过程，也让安全、伦理与用户体验进入作品评价。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "zarrow-generative-ai-studio-20260729", "limits": { "en": "Practitioner pilot account without comparative outcome evidence. Uses the visible byline and citation date, 29 July 2026; structured datePublished is 20 July and dateModified 29 July, so the original publication date is discrepant.", "zh": "教师对试点课程的观察，无对照效果研究。日期采用署名与引用行的2026-07-29；结构化datePublished为2026-07-20、dateModified为2026-07-29，原始发布日存在分歧。" }, "publishedAt": "2026-07-29", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Laura Zarrow · Wharton Generative AI Labs", "scope": { "en": "Read the public article’s teaching sequence, criteria and project cases; linked report and student work not independently assessed.", "zh": "阅读公开文章的教学流程、评价标准与项目案例；未独立评估学生作品或附件报告。" }, "status": "verified", "url": "https://gail.wharton.upenn.edu/research-and-insights/enduring-methods-emerging-medium/" } ], "stages": [ "higher" ], "summary": { "en": "Laura Zarrow describes a studio course using critique and timed design exercises to develop judgment through making.", "zh": "Laura Zarrow 复盘一门 AI 工作室课程：先发展想法，再制作，通过共同评议不断修改。" }, "title": { "en": "Enduring Methods for an Emerging Medium: The Generative AI Studio", "zh": "用作品讨论与限时设计，培养 AI 创作判断力" }, "topics": [ "teaching", "learning", "assessment" ], "type": "practice" }, { "body": { "en": [ "Watkins builds annotation, mapping, discussion and illustrated experiences around Faulkner without requiring student chatbot use. He reports weeks of development and multiple paid tools, while acknowledging that the result might become another distraction.", "The useful design test is which features prompt students to notice and interpret, rather than consume a substitute for reading." ], "zh": [ "Watkins 为福克纳小说制作批注阅读器、地图、讨论活动和插图版本，学生端不要求使用聊天机器人。他说明项目耗时数周、需要多个付费工具，并担心新体验也可能增加分心。", "可借鉴的不是复制全部媒体，而是明确设计约束：哪些功能会让学生停下来观察，哪些只是替他们完成阅读？" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "watkins-reading-machines-20260722", "limits": { "en": "Site not tested; no learning-effect evaluation reported.", "zh": "未运行网站；文章没有学习效果评估。" }, "publishedAt": "2026-07-22", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Marc Watkins · Rhetorica", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://marcwatkins.substack.com/p/reading-in-an-age-of-thinking-machines" } ], "stages": [ "higher" ], "summary": { "en": "A documented design experiment with open uncertainties.", "zh": "一个公开设计过程，也是一项尚未验证的尝试。" }, "title": { "en": "Reading In An Age Of Thinking Machines", "zh": "用 AI 建阅读环境，让学生自己读福克纳" }, "topics": [ "teaching", "learning", "tools" ], "type": "practice" }, { "body": { "en": [ "Sungu, Lira and Duckworth report that randomized teacher access to generative AI in Turkish schools reduced student intrinsic motivation by 0.11 SD, with no statistically significant average achievement effect reported. Their abstract also reports adverse results among lower-performing teachers.", "These preliminary abstract-level findings warrant checking student experience alongside workload. They do not establish mechanisms or generalize to every teacher or planning design." ], "zh": [ "Sungu、Lira 与 Duckworth 的工作论文摘要报告，在土耳其中学教师随机获得生成式 AI 支持后，学生内在动机低 0.11 标准差，平均成绩未显著改变。作者还报告较低表现教师的学生成绩与信心下降。", "这些是摘要层面的初步结果，尚不能用来确认机制、覆盖所有教师或否定特定备课方案。它的重要性在于要求同时测量减负、学习和学生体验。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "sungu-ai-teaching-rct-20260720", "limits": { "en": "Original indexed abstract and posting date only; full text returned 403. Methods unaudited; peer review not established.", "zh": "仅核验原站索引摘要和发布日期；全文 403，未读方法与附录；工作论文未确认同行评审。" }, "publishedAt": "2026-07-20", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Sungu, Lira & Duckworth / SSRN", "scope": { "en": "Original-site search index exposes abstract, authors, Posted 20 Jul 2026 and Date Written 25 Jun 2026; direct open returned 403.", "zh": "原站搜索索引显示完整摘要、作者、Posted 20 Jul 2026、Date Written 25 Jun 2026；直接打开 403。" }, "status": "limited", "url": "https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7007339" }, { "checkedAt": "2026-10-07T09:44:18Z", "name": "Alp Sungu official research page", "scope": { "en": "Official research page links the same SSRN paper; confirms attribution only.", "zh": "官方个人研究目录链接同一 SSRN 论文；仅核实归属。" }, "status": "verified", "url": "https://www.alpsungu.com/" } ], "stages": [ "school" ], "summary": { "en": "A recent working paper examines the tension between teacher productivity and student experience.", "zh": "近期工作论文把教师生产率与学生体验之间的张力带入实地研究。" }, "title": { "en": "Generative AI Can Harm Teaching", "zh": "教师获得 AI 工具后，学生动机未必改善" }, "topics": [ "teaching", "learning" ], "type": "research" }, { "body": { "en": [ "Warner frames writing as skills, knowledge and habits of mind. Using reported experiences with medical notes, he questions whether apparently clerical work also develops judgment. Similar artifacts may conceal different learning processes.", "For any time-saving proposal, identify the cognitive function of the removed step before deciding how success will be measured." ], "zh": [ "Warner 把写作视为技能、知识和思维习惯的结合。他借医学记录的报道提出疑问：看似可省掉的记录过程，是否本来就在帮助专业人员整理判断？相似成品未必意味着相同学习。", "适合审视任何省时方案：先列出被取消步骤原本承担的认知作用，再决定哪些成果指标能够发现损失。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "warner-outsourcing-real-work-20260709", "limits": { "en": "Medical example is reported secondhand, not clinical or learning evidence.", "zh": "医学例子来自他人报道；不是临床或学习效果证据。" }, "publishedAt": "2026-07-09", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "John Warner · Inside Higher Ed / Just Visiting", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://www.insidehighered.com/opinion/columns/just-visiting/2026/07/09/ai-scribes-medicine-are-short-circuiting-thinking" } ], "stages": [ "higher" ], "summary": { "en": "Automation can alter the practice behind an artifact.", "zh": "外包写作步骤，也可能改变专业思考。" }, "title": { "en": "What if We Accidentally Outsource the Real Work to AI?", "zh": "我们是否误把形成判断的练习当成杂务？" }, "topics": [ "teaching", "learning", "theory" ], "type": "opinion" }, { "body": { "en": [ "Watkins proposes combining focused and connected learning while examining online access, absences, assistive technologies, revision time and staff labor. He warns against turning instructor preferences into universal rules.", "Can students compare learning modes and develop judgment? If protected spaces are available only to privileged groups, resource allocation belongs in the pedagogical discussion." ], "zh": [ "Watkins 提议让专注活动与开放数字活动互相补充，同时逐项列出线上学生、缺课、辅助技术、课外修改及教师劳动的限制。他警惕把教师个人偏好直接变成所有学生必须遵守的规则。", "关键问题是学生能否比较不同学习方式，形成判断。若低技术体验只有少数班级负担得起，就需要一并讨论资源分配。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "watkins-analog-sustainability-20260708", "limits": { "en": "Design argument; several classroom examples are reported secondhand.", "zh": "设计论证；课堂案例多为作者转述。" }, "publishedAt": "2026-07-08", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Marc Watkins · Rhetorica", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://marcwatkins.substack.com/p/analog-is-having-a-moment-but-is" } ], "stages": [ "higher" ], "summary": { "en": "Testing the costs and access limits of analog responses.", "zh": "把线下专注空间当成设计问题，而非万能药。" }, "title": { "en": "Analog Is Having a Moment. But Is It Sustainable?", "zh": "低干扰课堂能否成为可持续的教学选择？" }, "topics": [ "teaching", "policy", "learning" ], "type": "opinion" }, { "body": { "en": [ "Evidence includes 26 students, 74 staff and analysis of 173 review reports. Inconsistent implementation emerges as a central concern.", "Recommendations emphasise substantive training, student participation and intelligible explanations of legitimate disciplinary differences." ], "zh": [ "原始材料包括26名学生、74名教职员工及173份审查报告；报告发现不同课程的执行与支持可能彼此冲突。", "它建议加强培训、从一开始纳入学生意见，并清楚解释合理的学科差异。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "qaa-perfect-storm-20260707", "limits": { "en": "Self-selected participants; no explicitly anti-GenAI staff participated. Not a nationally representative survey.", "zh": "参加者自选，未直接听到明确反对GenAI的员工；不能视为全英国代表性调查。" }, "publishedAt": "2026-07-07", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "QAA", "scope": { "en": "Read landing page and PDF findings, recommendations, methods and limitations; cited surveys not re-audited.", "zh": "阅读官方发布页与PDF主要发现、建议、技术方法和局限；未重读被引用调查。" }, "status": "verified", "url": "https://www.qaa.ac.uk/news-events/policy-and-leadership/state-of-the-nation" } ], "stages": [ "higher" ], "summary": { "en": "QAA examines the gap between institutional AI policies and students’ day-to-day assessment experience.", "zh": "QAA 的访谈和审查报告分析，把政策与学生实际经历之间的差距放在中心。" }, "title": { "en": "The perfect storm: AI, assessment and a sector under pressure", "zh": "AI 评价的制度风暴：同一学位内部的规则不一致" }, "topics": [ "assessment", "policy" ], "type": "research" }, { "body": { "en": [ "The episode connects understanding with building models. Its concluding reflection questions whether smoothing unfamiliar ideas into familiar forms can reduce the change involved in learning.", "A useful question is how assistance preserves manageable difficulty. The host also distinguishes such difficulty from material that is simply inaccessible to a learner." ], "zh": [ "节目将理解比作在已有心智结构中建模。末段的重点是：新材料若被过早改写成完全熟悉的样子，读者可能舒服地读完，却很少改变。", "这个观点适合与 AI 辅导一起讨论：怎样保留可承受的陌生感？节目也明确提醒，难到完全无法进入的材料并不因此更有学习价值。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "shishufeng-understanding-20260702", "limits": { "en": "Analogy and personal reading observations, not a summary-effect experiment. Transcript may contain errors; Hinton’s talk not independently reviewed.", "zh": "哲学类比与个人阅读观察；不是关于摘要效果的实验。官方转录可能有识别误差，Hinton 原演讲未独立核读。" }, "publishedAt": "2026-07-02", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "重轻 · 诗梳风 EP23", "scope": { "en": "Recovered official transcript; closely read the understanding argument and learning discussion from about 47:29 to the end. No audio playback. The publisher identifies this as an audio-derived transcript and warns of typographical/recognition errors; audio was not checked.", "zh": "恢复官方逐字稿，细读理解论证及约47:29至结尾的学习讨论；未听音频。 官方说明逐字稿整理自原始音频，可能有错别字或识别误差；本轮未核听音频。" }, "status": "limited", "url": "https://shishufeng.com/episode/what_is_understanding/transcript" } ], "stages": [ "general", "higher" ], "summary": { "en": "Chongqing uses Hinton’s account of understanding to question what learners lose when AI explanations become too smooth.", "zh": "重轻从 Hinton 的理解模型出发，反思 AI 摘要过于顺滑时，学习者可能失去的认知摩擦。" }, "title": { "en": "What Is Understanding? — Shishufeng EP23", "zh": "解释越顺，理解就越深吗？" }, "topics": [ "learning", "theory" ], "type": "opinion" }, { "body": { "en": [ "The 2023/24 task involved critiquing a generated grant proposal. Thirty of 230 students answered the survey; seven joined a focus group.", "Most respondents opposed AI integration into every assessment. Retrospective grade comparisons do not establish causal learning effects." ], "zh": [ "2023/24课程让学生评析ChatGPT 3.5生成的研究提案；230人中30人回应问卷，7人参加焦点组。", "多数回应者不赞成所有作业都使用 AI。历年成绩比较不能单独说明改革导致学习提升或没有损害。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "tam-student-ai-assessment-20260629", "limits": { "en": "13% response rate, one discipline and possible selection bias; read the repository accepted manuscript.", "zh": "13%回应率、单学科且可能自选偏差；当前阅读为机构存档接受稿。" }, "publishedAt": "2026-06-29", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Angela Tam, David Abraham & Andrew Williams · Intersection / UCL Discovery", "scope": { "en": "Read accepted-manuscript design, methods, sample, results, discussion and limitations; appendices not audited.", "zh": "阅读接受稿设计、方法、样本、结果、讨论及局限；附录未逐项审计。" }, "status": "verified", "url": "https://discovery.ucl.ac.uk/id/eprint/10228028/" } ], "stages": [ "higher" ], "summary": { "en": "Medical-science students see value in critiquing AI but resist universal AI integration.", "zh": "医学科学学生既看到评价 AI 输出的价值，也反对把 AI 加入所有作业。" }, "title": { "en": "Integrating Generative AI Into University Coursework Assessments – a Student Perspective", "zh": "学生如何看待批评 AI 科学提案的作业？" }, "topics": [ "assessment", "learning" ], "type": "research" }, { "body": { "en": [ "They accept targeted controls while arguing for layered evidence across a program.", "Their central trade-off is access: returning to examination centres may exclude learners without resolving the underlying quality of assessment design." ], "zh": [ "他们支持必要的受控验证，但反对把所有线上评价问题归结为教学方式。", "建议跨课程累积多种证据，并明确投入设计、促进学习及人力支持，而不是把线上规模误认为低成本。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "smith-online-modality-risk-20260628", "limits": { "en": "Practitioner argument, not statistical proof that modalities carry identical risks.", "zh": "多位实践者的立场文；不是线上线下风险相等的统计证明。" }, "publishedAt": "2026-06-28", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Liz Smith, Dawn Gilmore, Chinh Nguyen, Paul Wiseman, Caitlin Yolland & Jim Frederickson · Needed Now", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://needednowlt.substack.com/p/modality-does-not-equal-risk" } ], "stages": [ "higher" ], "summary": { "en": "Online-education practitioners challenge the assumption that physical attendance alone assures learning.", "zh": "在线教育实践者提醒，回到现场考点会影响可及性，也不能代替完整的评价设计。" }, "title": { "en": "Modality does not equal risk", "zh": "线上教学方式不等于评价风险" }, "topics": [ "assessment", "policy" ], "type": "opinion" }, { "body": { "en": [ "TEQSA’s commissioned report connects adaptive capabilities with disciplinary knowledge and learning-process evidence. It proposes process-focused assessment and support for student judgment, agency and self-regulation." ], "zh": [ "TEQSA 委托报告把数字素养、分布式认知、混合元认知与终身学习作为适应性能力，强调学科知识基础。它建议高校收集学习过程证据，调整评价设计，支持学生判断何时依赖或质疑 AI。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "teqsa-adaptive-capabilities-20260624", "limits": { "en": "Evidence-informed expert recommendations, not a causal intervention study. The report is explicitly not TEQSA Guidance Notes or a uniform mandate.", "zh": "证据支持的专家建议，不是教学方案的因果效果试验；报告明确不是 TEQSA Guidance Notes，也不作统一强制规定。" }, "publishedAt": "2026-06-24", "sources": [ { "checkedAt": "2026-10-07T07:32:44.119715+00:00", "name": "TEQSA", "scope": { "en": "Official introduction and 28-page public PDF read, focusing on purpose, definitions and recommendations. Publication date comes from the official recent-resources index. A local link recheck on 7 October 2026 returned a 15-second timeout; earlier research-reader access succeeded. Access differs by route.", "zh": "已读官方介绍和28页公开PDF，重点核读目的、能力定义与建议。发布日期来自官方近期资源索引。 本地链接复测（2026-10-07）返回15秒超时；此前研究读取可读，不代表所有访问路径都畅通。" }, "status": "limited", "url": "https://www.teqsa.gov.au/guides-resources/resources/corporate-publications/assuring-quality-learning-gen-ai-integrated-future-role-adaptive-capabilities" } ], "stages": [ "higher" ], "summary": { "en": "TEQSA’s commissioned report connects adaptive capabilities with disciplinary knowledge and learning-process evidence. It proposes process-focused assessment and support for student judgment, agency and self-regulation.", "zh": "TEQSA 委托报告把数字素养、分布式认知、混合元认知与终身学习作为适应性能力，强调学科知识基础。它建议高校收集学习过程证据，调整评价设计，支持学生判断何时依赖或质疑 AI。" }, "title": { "en": "Assuring quality learning in a gen AI-integrated future: adaptive capabilities", "zh": "TEQSA：让学习过程成为 AI 时代的质量证据" }, "topics": [ "assessment", "learning", "policy" ], "type": "policy" }, { "body": { "en": [ "The prototype maps outcomes to open and verified tasks, exposing gaps and clustering across a degree.", "It makes design information inspectable for governance; mapping an assessment is still different from validating an individual student’s capability." ], "zh": [ "PowerBI 原型把课程与学位目标、开放及验证任务连起来，便于看出某个目标是否缺乏证据或验证过度集中。", "它解决的是治理团队看不见已有设计资料的问题，并不等于仪表盘上的映射已经证明每位学生具备能力。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "unisq-assurance-dashboard-20260621", "limits": { "en": "Explicitly a prototype; not run here and no learning-effect or validity trial reported.", "zh": "作者明确称仍是原型；未运行系统，也没有学生学习效果或效度试验。" }, "publishedAt": "2026-06-21", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Jessica Marrington, Chris Zehntner, Neil Martin & Jo-Anne Ferreira · Needed Now", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://needednowlt.substack.com/p/what-does-assurance-of-learning-actually" } ], "stages": [ "higher" ], "summary": { "en": "UniSQ’s team describes a prototype connecting program outcomes, tasks and assessment security.", "zh": "UniSQ 团队展示把学习目标、任务及验证环节放进同一视图的原型。" }, "title": { "en": "What does assurance of learning actually look like in practice?", "zh": "把整项学位的学习保证变得可见" }, "topics": [ "assessment", "policy", "tools" ], "type": "practice" }, { "body": { "en": [ "A Year 9 teacher made 26 comparisons and dictated comments on 26 scripts, combining them with AI judging. Logged judging and comments took under 20 minutes; pupils valued knowing their teacher endorsed feedback.", "The four-hour comparison is the author’s estimate, not a measured control. Familiarity with pupils and drafts matters, and the timing does not capture the entire instructional workflow." ], "zh": [ "一位九年级英语教师对 26 份作文完成 26 次人类比较判断并口述反馈；系统再结合 AI 判断。记录显示判断加口述不足 20 分钟，学生强调需要知道教师认可 AI 意见。", "标题的“四小时”是作者估计的传统批改耗时，并非同一教师的实测对照。教师早已熟悉学生及草稿；课前教学、准备和后续反馈课不能算进零成本。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-classroom-marking-case-20260620", "limits": { "en": "Single-class vendor-reported experience; no independent learning test or full-workload comparison.", "zh": "单班、厂商转述教师经验；没有独立学习测验或完整工时对照。" }, "publishedAt": "2026-06-20", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/an-ai-marking-case-study-from-four" } ], "stages": [ "school" ], "summary": { "en": "A classroom case exposes teacher reading, student trust and ambiguous AI feedback.", "zh": "具体课堂案例保留了教师阅读、学生信任和 AI 模糊反馈的问题。" }, "title": { "en": "An AI marking case study: from four hours to twenty minutes", "zh": "26 名学生的混合批改：时间账如何计算" }, "topics": [ "teaching", "assessment" ], "type": "practice" }, { "body": { "en": [ "Their personal stories distinguish liking an activity from being good at it, and emphasize trying meaningful work and sustaining motivation.", "The episode offers a question about agency and educational purpose. Its creators’ unusual careers and broad labor-market claims limit its usefulness as concrete admissions or career advice." ], "zh": [ "两位从自身经历讨论：喜欢与擅长未必相同，学校之外的实践也能帮助人发现方向。节目把注意力转向持续参与真实的事情、形成连接和保护学习动机。", "值得读的是教育目的与个人能动性的追问。主持人的职业处境有明显选择性，关于学历回报和未来经济的判断不能直接用作升学建议。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "shishufeng-exam-life-paths-20260615", "limits": { "en": "Strong personal views with selection and survivorship limitations. Demographic, labor and credential-return claims not verified; English title is translated.", "zh": "个人经验与强烈观点，带有幸存者和职业样本局限；未核实人口、就业与学历收益论断。英文题为译题。" }, "publishedAt": "2026-06-15", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "汉洋、重轻 · 诗梳风 EP22", "scope": { "en": "Recovered official transcript; close-read education choices, motivation, practical exploration and concluding reflections. Audio not played. The publisher identifies this as an audio-derived transcript and warns of typographical/recognition errors; audio was not checked.", "zh": "恢复官方逐字稿，重点阅读升学经历、动机、实践及结尾反思；未听音频。 官方说明逐字稿整理自原始音频，可能有错别字或识别误差；本轮未核听音频。" }, "status": "limited", "url": "https://shishufeng.com/episode/gaokao/transcript" } ], "stages": [ "school", "family", "higher" ], "summary": { "en": "Hanyang and Chongqing use university choices to question the idea that one decision fixes an entire life path.", "zh": "汉洋与重轻借选专业谈学习动机、真实尝试与人生路径，质疑一次选择可以安排一生的想象。" }, "title": { "en": "Just an Exam — Shishufeng EP22", "zh": "一次考试，能替整个人生定轨吗？" }, "topics": [ "theory", "learning" ], "type": "opinion" }, { "body": { "en": [ "She abandoned oral exams when inexperienced instructors struggled with consistency, and replaced take-home work with peer review and in-class group quizzes.", "Tutor-signature requirements overwhelmed support capacity. Simplification helped workload, but unsuccessful outcomes remained an unresolved problem." ], "zh": [ "因新教师难以一致实施，她放弃口试；因AI生成作业问题，改为同伴评阅后的小组现场测验。", "重考前辅导一度压垮支持中心，课程最终靠简化标准与流程缓解负担；低分和退课问题仍未消失。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "freeman-grading-six-years-20260615", "limits": { "en": "Longitudinal practitioner account, not randomised evidence or a verified cheating-rate estimate.", "zh": "课程协调者长期自述；观察到的表现变化不是随机对照结果，也没有可靠作弊率。" }, "publishedAt": "2026-06-15", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Hilary Freeman · Grading for Growth", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://gradingforgrowth.com/p/standards-based-grading-six-years" } ], "stages": [ "higher" ], "summary": { "en": "Freeman revisits a coordinated calculus course of 250–450 students and the compromises required to sustain it.", "zh": "Freeman 复盘250至450人的微积分课程，公开口试、课外作业和辅导签字安排的失效点。" }, "title": { "en": "Standards-Based Grading Six Years Later", "zh": "六年后的标准本位评分：大班课程真实改了什么？" }, "topics": [ "assessment", "teaching" ], "type": "practice" }, { "body": { "en": [ "Christodoulou accepts adaptation based on demonstrated performance but questions organizing content around preferences. Analogies must be familiar and illuminate the target concept; entertaining contexts can also distract.", "The essay separates pacing, remediation, interest-based rewriting and learner choice. Its provocative title should not be treated as evidence against every form of adaptive instruction." ], "zh": [ "Christodoulou 认可依据答题表现安排后续练习，却质疑由学生偏好决定内容与学习方式。类比只有在学生熟悉、并真正解释目标概念时才有帮助；有趣也可能分散注意。", "这将“个性化”拆成不同假设：调整速度、补缺、换背景和让学生自由选择并不是同一种干预。文章鼓励逐例验证，不应把其强标题当作已否定所有适应性教学。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-personalization-interests-20260613", "limits": { "en": "Mechanism argument and teaching examples, not a head-to-head trial of personalization approaches.", "zh": "机制论证与教学例子；不是各种个性化方案的直接比较试验。" }, "publishedAt": "2026-06-13", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/why-personalising-education-based" } ], "stages": [ "school", "family" ], "summary": { "en": "Performance-based adaptation is distinguished from rewriting every topic around a learner’s interests.", "zh": "作者支持按表现调整练习进度，同时质疑把所有知识改写成学生的兴趣。" }, "title": { "en": "Why personalising education based on student interests doesn’t work", "zh": "按兴趣个性化，不等于按学习需要适配" }, "topics": [ "learning", "theory", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Nearly 9,000 Year 6 pupils in 211 schools wrote independently, then revised after mixed human–AI feedback. The company reports an eight-point gain, about 0.2 effect size; 64% scored higher.", "The important questions concern understanding and acting on feedback. Improvement in a revised artifact neither isolates AI’s contribution nor demonstrates independent transfer." ], "zh": [ "近 9,000 名六年级学生、211 所学校先独立写作，数周后收到人类与 AI 混合反馈并修改同一篇作品。公司报告平均提高 8 个量尺分、约 0.2 效应量，64% 学生第二次得分更高。", "关键不是生成多少评语，而是学生是否理解、行动并改变以后写作。这里测到的是一次修改后成品变化，不能单独归因于 AI，也不能视为独立迁移。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-redraft-feedback-20260530", "limits": { "en": "Vendor pre/post report without randomization; mixed feedback; no delayed, novel-task, unaided assessment.", "zh": "厂商前后测，无随机对照；人机混合反馈；没有延迟、新题、无 AI 测验。" }, "publishedAt": "2026-05-30", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/using-ai-feedback-to-improve-writing" } ], "stages": [ "school" ], "summary": { "en": "A large redrafting project makes uptake of feedback visible while leaving learning unresolved.", "zh": "大型前后改稿项目让反馈是否被理解和采用成为可追问的问题。" }, "title": { "en": "Is AI feedback true but useless?", "zh": "九千名小学生改稿：成品进步，学习仍待检验" }, "topics": [ "assessment", "learning", "teaching" ], "type": "practice" }, { "body": { "en": [ "Mollick contrasts answer delivery with support that makes learners solve problems. He also describes protecting his writing practice while using AI for checks and reader perspectives.", "The practical question is what to delegate and what to keep practicing. His research examples motivate that choice, but do not establish one universal effect of AI use." ], "zh": [ "作者把替人完成作业与推动人自己解题区分开，也说明自己如何保留写作练习、把 AI 用作读者和检查者。", "他的核心问题是：哪些工作可以交出去，哪些能力值得亲自练习？文中研究是其论证依据；不同研究设计的差异不能被压成一个通用效果。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "mollick-choosing-stay-human-20260526", "limits": { "en": "Author interpretation and personal practice. School-year conversions and product steps are not adopted as verified guidance.", "zh": "作者的研究解读与个人实践；不复述文中学年等值换算，也不将其产品用法当作当前操作指南。" }, "publishedAt": "2026-05-26", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Ethan Mollick · One Useful Thing", "scope": { "en": "Read the full public essay; cited studies and product instructions not independently re-evaluated.", "zh": "阅读全文；未独立重做所引研究或检查所有产品操作步骤。" }, "status": "verified", "url": "https://www.oneusefulthing.org/p/choosing-to-stay-human" } ], "stages": [ "general", "higher", "school" ], "summary": { "en": "Mollick considers how convenient AI defaults can bypass the effort through which people develop valued abilities.", "zh": "Mollick 从写作与学习讨论 AI 默认设置：方便交出成果，也可能让人跳过形成能力的过程。" }, "title": { "en": "Choosing to Stay Human", "zh": "哪些思考值得自己保留？" }, "topics": [ "learning", "teaching", "theory" ], "type": "opinion" }, { "body": { "en": [ "After a week with a social-media detector extension, Watkins reports attending to labels rather than arguments and observing inconsistent classifications across contexts. He favors deliberate disclosure and process explanations.", "Read against Mills’s limited use of detection: distinguish classifier accuracy, changes in the reader’s behavior, and opportunities to contest a label." ], "zh": [ "Watkins 回顾使用社交媒体 AI 检测扩展的一周：他开始先看标签，再决定是否阅读，并记录同一内容在不同上下文中分类变化。他因此更重视主动披露与过程解释。", "与 Mills 的有限检测立场对读，可把争论拆成准确性、使用者行为和被标记者申辩机会三个问题。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "watkins-detector-trust-20260503", "limits": { "en": "Personal trial, not a replicated accuracy evaluation.", "zh": "个人测试；既未复现实验，也不采信厂商准确率。" }, "publishedAt": "2026-05-03", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Marc Watkins · Rhetorica", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://marcwatkins.substack.com/p/how-an-ai-detector-made-me-trust" } ], "stages": [ "higher" ], "summary": { "en": "A personal trial raises questions about labels and trust.", "zh": "一周亲测，引出标签、信任与正当程序。" }, "title": { "en": "How An AI Detector Made Me Trust People Less", "zh": "检测标签如何改变教师自己的阅读判断" }, "topics": [ "assessment", "tools", "theory" ], "type": "opinion" }, { "body": { "en": [ "Christodoulou and Wheadon distinguish reliability from validity, discussing human agreement, repeated-assessment correlations and developmental patterns. They retain human reading and judgment.", "They identify unfinished work on unusual responses, external tests, adversarial gaming and instructional washback. Their inability to find a gaming strategy is not equivalent to systematic adversarial validation." ], "zh": [ "Christodoulou 与 Wheadon 区分评分可靠性和效度，讨论人机一致、跨次测验相关及年龄增长曲线。他们认为现有结果有希望，同时保留人类阅读与判断。", "作者明确列出未解决的问题：异常作品、外部测验关联、对抗性操纵，以及教师是否为迎合 AI 而改变教学。尚未发现作弊办法，并不等于经过系统攻防检验。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-validity-black-box-20260425", "limits": { "en": "Vendor validity argument; raw data, bias and adversarial robustness were not independently tested.", "zh": "厂商效度论证；未取得原始数据，也未独立检验偏差和可操纵性。" }, "publishedAt": "2026-04-25", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou & Chris Wheadon / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/the-black-box-of-ai-assessment" } ], "stages": [ "school" ], "summary": { "en": "A vendor’s validity review recognizes that stable scores can still reward the wrong learning.", "zh": "厂商自己的效度复盘承认，稳定一致的评分也可能激励错误的教学方向。" }, "title": { "en": "The black box of AI assessment", "zh": "AI 评分稳定，仍需证明它评的是目标能力" }, "topics": [ "assessment", "theory", "tools" ], "type": "vendor" }, { "body": { "en": [ "The argument focuses on separability and observability, not simply another cheating device.", "It asks whether escalating inspection shifts burdens onto students with disabilities, health needs or religious dress." ], "zh": [ "论文以可分离性与可观察性为核心，认为普通眼镜、助听器等与 AI 融合会使排除规则更难实施。", "它也指出身体检查及设备判定的成本可能不均匀地落在有残障、健康需要或宗教服饰的学生身上。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "corbin-wearable-ai-assessment-20260424", "limits": { "en": "Conceptual analysis read through publisher-indexed abstract and main sections; no device test or cheating-prevalence study.", "zh": "概念分析；阅读出版社索引中的摘要与主要段落，非实地设备试验或作弊率调查。" }, "publishedAt": "2026-04-24", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Thomas Corbin, Sue Sharpe & Phillip Dawson", "scope": { "en": "Read publisher-indexed abstract, main argument and conclusion; device examples not independently checked.", "zh": "阅读出版社索引的摘要、主要论证及结论；未核验全部设备例子。" }, "status": "limited", "url": "https://doi.org/10.1080/02602938.2026.2661367" } ], "stages": [ "higher" ], "summary": { "en": "The authors question assessment designs that depend on detecting and excluding wearable assistance.", "zh": "Corbin、Sharpe 与 Dawson 追问：如果辅助工具难以看见或摘除，现场评价的保障来自哪里？" }, "title": { "en": "On AI glasses and wearable AI in assessment", "zh": "可穿戴 AI 对“隔离工具”的评价设计提出什么挑战？" }, "topics": [ "assessment", "theory", "policy" ], "type": "opinion" }, { "body": { "en": [ "Students prepare published proofs, reconstruct them unaided, receive immediate feedback and can retry on later days.", "Clark explicitly excludes creativity as the target. This is assessment-design background for AI-era debates, not an AI intervention." ], "zh": [ "实分析学生先准备公开题目，再在不使用资料的情况下重建证明，教师现场反馈，并允许不同日期再次尝试。", "作者明确说这不适合把创造性作为目标的任务；小班预约负担与口头反馈条件也限制推广。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "clark-mastery-exams-20260413", "limits": { "en": "Classes capped at 24; anxiety is self-reported, and the separation of memorisation from understanding is not independently validated.", "zh": "每班上限24人的教师复盘；焦虑来自学生自报，未验证记忆与理解能被完全区分。" }, "publishedAt": "2026-04-13", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "David Clark · Grading for Growth", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://gradingforgrowth.com/p/mastery-exams" } ], "stages": [ "higher" ], "summary": { "en": "Clark uses known proof tasks and flexible timing to distinguish understanding from spontaneous invention.", "zh": "Clark 让学生预知证明题并选择应试时间，考查能否重建逻辑而不是现场创造。" }, "title": { "en": "Mastery Exams", "zh": "掌握型考试：让限时验证不必等于临场发明" }, "topics": [ "assessment", "teaching" ], "type": "practice" }, { "body": { "en": [ "The five-department action plan connects AI literacy, university curricula, teacher development, applications and safeguards, with goals for 2030. These are policy directions, not completed implementation or demonstrated learning effects." ], "zh": [ "五部门行动计划连接中小学 AI 通识、高校课程、教师智能素养、教学科研应用与安全治理，并提出2030年目标。文件提出的智能学伴、资源和评估应用属于建设方向，不能视为已完成或验证有效。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "china-ai-education-action-plan-20260410", "limits": { "en": "Policy goals and requirements; local and school implementation needs separate checking. The date shown is official webpage publication, 10 April 2026; the document is dated 2 April 2026 (教科信〔2026〕1号).", "zh": "政策目标与实施要求，具体安排需核对地方和学校文件。此处日期为官方网页发布日2026年4月10日；文件落款为4月2日，教科信〔2026〕1号。" }, "publishedAt": "2026-04-10", "sources": [ { "checkedAt": "2026-10-07T07:32:44.134039+00:00", "name": "教育部等五部门 / 国家数据局", "scope": { "en": "Full text read on the National Data Administration website, a co-issuing authority. The Ministry of Education entry redirected or timed out during this check.", "zh": "已读共同发文机构国家数据局官网全文；教育部入口本次出现重定向/超时，未以此推定所有官方入口失效。" }, "status": "verified", "url": "https://www.nda.gov.cn/sjj/zwgk/tzgg/0410/20260410162428109058250_pc.html" } ], "stages": [ "school", "higher", "general" ], "summary": { "en": "The five-department action plan connects AI literacy, university curricula, teacher development, applications and safeguards, with goals for 2030. These are policy directions, not completed implementation or demonstrated learning effects.", "zh": "五部门行动计划连接中小学 AI 通识、高校课程、教师智能素养、教学科研应用与安全治理，并提出2030年目标。文件提出的智能学伴、资源和评估应用属于建设方向，不能视为已完成或验证有效。" }, "title": { "en": "China’s AI+Education Action Plan: curriculum, teacher development and governance", "zh": "中国“人工智能+教育”行动计划：课程、师资与治理同步推进" }, "topics": [ "policy", "teaching", "tools" ], "type": "policy" }, { "body": { "en": [ "A polished artifact no longer automatically evidences the process educators hope it represents.", "The proposed rubric audit asks which criteria reveal understanding, quality judgment and transfer rather than surface fluency." ], "zh": [ "一份流畅成品不能自动证明学生经历过学习；即使分成开放与封闭两轨，开放作业的解释问题仍在。", "文章提出检查现有标准：哪些只需表面精良的 AI 输出就能满足，哪些真正要求学生展示理解。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "tifflin-marking-now-20260406", "limits": { "en": "Design argument without an intervention evaluation; it does not demonstrate independent learning.", "zh": "设计主张，未报告试验；不能据此宣称新标准已证明独立学习。" }, "publishedAt": "2026-04-06", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Raelene Tifflin · Needed Now", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://needednowlt.substack.com/p/what-are-we-going-to-do-about-marking" } ], "stages": [ "higher" ], "summary": { "en": "Tifflin asks markers to make conceptual grasp, evaluative judgment and transfer visible.", "zh": "Tifflin 建议把概念理解、质量判断与情境迁移直接写入设计与评分标准。" }, "title": { "en": "What are we going to do about marking now?", "zh": "允许 AI 的作业应该评什么？" }, "topics": [ "assessment", "theory" ], "type": "opinion" }, { "body": { "en": [ "Repeated grading trials compare clean papers with simple embedded attacks in student and research-paper corpora. Most frontier-model conditions show small changes, with model-specific exceptions.", "Models rarely announce the attacks even when they resist them. For educators, resistance to manipulation and the validity or fairness of the underlying grade remain separate questions." ], "zh": [ "研究使用学生论文与研究论文，反复比较有无简单文本注入的评分。多数前沿模型在多数条件下变化较小，但部分模型和长文条件更脆弱。", "即使没有被明显操纵，模型通常也不会主动报告发现了注入。对教师而言，抵抗这类攻击与评分本身准确、公平，是需要分别检验的问题。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "gail-grading-prompt-injections-20260402", "limits": { "en": "Specific models, rubric, papers and simple plaintext attacks; not evidence that AI grading is generally secure or valid.", "zh": "限定于所测模型、评分规则、论文与简单明文攻击，不能解释为 AI 评分已经安全或有效。" }, "publishedAt": "2026-04-02", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Benjamin Wanjura, Dan Shapiro, Ethan Mollick, Lilach Mollick, Lennart Meincke · Wharton GAIL", "scope": { "en": "Read public report methods, results and limitations; not the SSRN attachment or every plotted value.", "zh": "阅读公开技术报告页的方法、结果与限制；未核读 SSRN 附件或逐项检查图中数值。" }, "status": "verified", "url": "https://gail.wharton.upenn.edu/research-and-insights/hidden-prompt-injections/" } ], "stages": [ "higher" ], "summary": { "en": "A Wharton team tests how embedded instructions affect AI grading, finding important differences across models and conditions.", "zh": "Wharton 团队测试隐藏指令能否抬高 AI 给论文的分数，发现模型、文本长度和放置位置都会影响风险。" }, "title": { "en": "This is an Excellent Paper: The Effects of Prompt Injection on Grading", "zh": "AI 评分抗干扰，要按模型和材料来检查" }, "topics": [ "assessment", "tools" ], "type": "research" }, { "body": { "en": [ "Conversations with teachers, pupils and parents surface speed, accuracy, human contact, deskilling and environmental concerns. Pupils reportedly value teachers reading their work while sometimes welcoming another perspective.", "The design questions include correcting errors and preserving teachers’ judgment practice. Without sampling details or prevalence estimates, this account cannot represent stakeholder opinion generally." ], "zh": [ "作者根据与教师、学生和家长的交流，归纳速度、准确性、人际接触、教师技能流失及环境影响五类问题。她强调学生在意教师亲自阅读作品，也有人希望 AI 提供不同视角。", "这使讨论超出节省多少批改时间：如何申诉错误、怎样保留教师练习判断的机会、哪些关系不能被自动评语替代？报告没有给出访谈抽样或发生率，不能当作群体民意调查。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-assessment-stakeholder-concerns-20260321", "limits": { "en": "Company-authored conversation summary, not a representative survey; environmental and exam-statistics claims not independently audited.", "zh": "公司人员总结交流，非代表性调查；本文环境影响及考试评分数字未独立复核。" }, "publishedAt": "2026-03-21", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/the-ethics-of-ai-assessment-five" } ], "stages": [ "school", "family" ], "summary": { "en": "Stakeholder conversations surface teacher deskilling and pupils’ wish to be read seriously.", "zh": "与教师、学生及家长的交流，让教师能力流失与被认真阅读成为设计问题。" }, "title": { "en": "The ethics of AI assessment: five big issues", "zh": "AI 批改的伦理：快、准之外还有师生关系" }, "topics": [ "assessment", "teaching", "policy" ], "type": "opinion" }, { "body": { "en": [ "Examples include editing shared styles page by page and requesting a rollback without saved versions. These reveal missing mental models of the working environment.", "The hosts emphasize learning through use. For adult AI literacy, an editorial implication is to observe where novices struggle before choosing which concepts to teach." ], "zh": [ "对话举出逐页修改共享样式、让 AI “改回去”却未保存版本等例子。问题常不在提示词花样，而在学习者尚未形成对任务环境的理解。", "他们强调通过实际操作形成手感。用于成人 AI 素养教学时，可先观察学习者卡在哪一步，再补上具体概念，而非假设人人具备相同基础。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "shishufeng-teaching-ai-novices-20260319", "limits": { "en": "Informal training observations, with time-specific tool claims; not evidence against structured courses. English title is translated.", "zh": "非正式培训观察；个别工具能力与说法有时间和情境限制，不能据此否定系统课程。英文题为译题。" }, "publishedAt": "2026-03-19", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "汉洋、重轻 · 诗梳风 EP9", "scope": { "en": "Recovered transcript; close-read training, tacit knowledge, structure and version-management cases. No audio or replication. The publisher identifies this as an audio-derived transcript and warns of typographical/recognition errors; audio was not checked.", "zh": "恢复官方逐字稿，细读培训动机、默会知识、结构与版本管理案例；未听音频或复现操作。 官方说明逐字稿整理自原始音频，可能有错别字或识别误差；本轮未核听音频。" }, "status": "limited", "url": "https://shishufeng.com/episode/AI_for_normies/transcript" } ], "stages": [ "general", "higher" ], "summary": { "en": "The hosts find that tacit knowledge about files, versions and structure often explains beginners’ difficulties.", "zh": "两位从培训朋友的经历发现：文件、版本和结构这些默认常识，常是新手真正卡住的地方。" }, "title": { "en": "Teaching Friends to Use AI — Shishufeng EP9", "zh": "教人用 AI，先看见那些没说出口的基础" }, "topics": [ "teaching", "learning", "tools" ], "type": "practice" }, { "body": { "en": [ "The July 26 v3 covers 73 students and engineering fixes for stacked questions and biased case selection; interruptions persisted.", "It distinguishes marginal grading costs from subscription and voice costs, and explicitly lacks blinded expert validation and subgroup fairness analysis." ], "zh": [ "本文初稿3月发布，本轮阅读7月26日v3：系统修复了部分连环提问与抽题偏差，但打断思考仍存在。", "作者区分订阅内评分边际费用与完整语音成本，并明确尚未做盲法专家评分验证或群体公平性分析。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "ipeirotis-rizakos-voice-oral-20260318", "limits": { "en": "Single-course cohorts with simultaneous changes; no causal attribution. The 2025 blog is linked, not counted again.", "zh": "单课程两届自述；多个改动同时发生，不能归因。2025博客并入同一研究链，不重复计数。" }, "publishedAt": "2026-03-18", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Panos Ipeirotis & Konstantinos Rizakos · arXiv / CACM", "scope": { "en": "Read v3 design, costs, reliability, failures, student experience and validity limitations; system not run.", "zh": "阅读v3设计、费用、可靠性、故障、学生体验及效度边界部分；未运行系统。" }, "status": "verified", "url": "https://arxiv.org/abs/2603.18221" }, { "checkedAt": "2026-10-07T09:47:56Z", "name": "Panos Ipeirotis · original classroom blog, 2025-12-29", "scope": { "en": "Read the original report and paper-expansion note; demonstration video not watched.", "zh": "阅读原始复盘及其论文扩展说明；未看演示视频。" }, "status": "verified", "url": "https://www.behind-the-enemy-lines.com/2025/12/fighting-fire-with-fire-scalable-oral.html" } ], "stages": [ "higher" ], "summary": { "en": "Two cohorts show operational feasibility, while agreement among model graders does not establish validity.", "zh": "两届73人的实践显示运行可行性，但模型彼此一致还不能证明评分正确。" }, "title": { "en": "Scalable and Personalized Oral Assessments Using Voice AI", "zh": "语音 AI 口试：两届实践、工程故障与评分效度边界" }, "topics": [ "assessment", "tools" ], "type": "practice" }, { "body": { "en": [ "A Stanford review identified 20 higher-quality causal studies within over 800 K12 AI papers available by October 2025. Assisted performance and unaided learning diverge; long-term, equity and wellbeing evidence remains limited." ], "zh": [ "Stanford 团队从截至2025年10月的800余篇 K12 AI 论文中识别20项较高质量因果研究。综述提醒：有 AI 帮助时任务表现提高，不等于撤去工具后能独立掌握；长期学习、公平与福祉仍有证据缺口。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "stanford-k12-evidence-review-20260311", "limits": { "en": "Bounded by the evidence database, inclusion criteria and October 2025 cutoff; not a review of every study available today.", "zh": "综述受检索库、纳入标准与2025年10月截止时间限定，不代表截至今天的所有研究。" }, "publishedAt": "2026-03-11", "sources": [ { "checkedAt": "2026-10-07T07:32:44.134252+00:00", "name": "Stanford SCALE Initiative / AI Hub for Education", "scope": { "en": "Full research-team explanation read. The report is accessible, but this note follows the explanation and does not claim independent checking of every included study.", "zh": "核读研究团队完整说明页。公开报告可访问，摘要依据说明页，未逐篇复核纳入研究。" }, "status": "verified", "url": "https://scale.stanford.edu/research-in-action/understanding-evidence-base-ai-k12-education" } ], "stages": [ "school" ], "summary": { "en": "A Stanford review identified 20 higher-quality causal studies within over 800 K12 AI papers available by October 2025. Assisted performance and unaided learning diverge; long-term, equity and wellbeing evidence remains limited.", "zh": "Stanford 团队从截至2025年10月的800余篇 K12 AI 论文中识别20项较高质量因果研究。综述提醒：有 AI 帮助时任务表现提高，不等于撤去工具后能独立掌握；长期学习、公平与福祉仍有证据缺口。" }, "title": { "en": "Understanding the Evidence Base on AI in K-12 Education", "zh": "Stanford K12 证据综述：完成任务与持久学习要分开看" }, "topics": [ "learning", "theory", "assessment" ], "type": "research" }, { "body": { "en": [ "Seitz describes in-class drafting, conversational feedback and specifications-style grading. Most reported reactions were positive, but a slower writer objected. He regretted moving short exercises to handwriting because output, readability and recordkeeping suffered.", "The design question is how to protect attention without imposing one writing rhythm. Device choice, learning environment and grading can be considered separately." ], "zh": [ "Seitz 用课堂短篇写作、对话式反馈和宽松的达标评价保护思考过程。学生反馈多为积极，但有人需要更多时间。他后悔把短练习改成手写：学生写得更少，自己阅读与记录更费力。", "此例适合讨论如何保护注意力，同时容纳不同写作节奏。保留电脑、改变任务环境和改变评分方式，是可以分别选择的设计。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "bruff-seitz-in-class-writing-20260310", "limits": { "en": "Self-report and course feedback, not causal evidence.", "zh": "自述和课程反馈；不是因果试验。" }, "publishedAt": "2026-03-10", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Derek Bruff · Intentional Teaching", "scope": { "en": "Read writing, grading and handwriting-reflection sections; audio unchecked.", "zh": "已读写作、评价与手写反思段落；未核听音频。" }, "status": "verified", "url": "https://intentionalteaching.buzzsprout.com/2069949/episodes/18794328-in-class-writing-with-james-seitz" } ], "stages": [ "higher" ], "summary": { "en": "Protected writing time, with candid design reversals.", "zh": "低干扰写作并不等于必须手写。" }, "title": { "en": "In-Class Writing with James Seitz", "zh": "把写作搬进课堂，也承认纯手写的失败" }, "topics": [ "teaching", "learning", "assessment" ], "type": "practice" }, { "body": { "en": [ "Drawing on Kant and other philosophers, Hui defends universities as places for collective inquiry, contesting reasons, and learning how to live. He questions their reorganization around industrial agendas.", "The essay opens a useful curriculum question: what forms of shared intellectual life matter even if information becomes abundant? Its value lies in a contestable account of educational purpose." ], "zh": [ "许煜借康德等思想资源，主张大学还应让人共同检验理由、争论知识，并学习怎样生活。他担心技术产业的议程把这些目的挤出大学。", "这篇文章值得与 AI 教学效率讨论并读：更快获得答案之后，哪些共同探究仍值得保留？这是对教育目的的哲学论证，其前提可以争辩。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "hui-university-automated-knowledge-20260304", "limits": { "en": "Normative philosophy. Automation scenarios are premises, not established outcomes; the Chinese heading is editorial.", "zh": "规范性哲学论证；有关未来自动化的设想不应当作已发生的事实。中文标题是策展译题。" }, "publishedAt": "2026-03-04", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Yuk Hui · e-flux journal #161", "scope": { "en": "Read all three main sections and conclusion; checked original date metadata, not every cited work.", "zh": "阅读三节主论证及结尾，核对原站日期元数据；注释所引著作未逐一核读。" }, "status": "verified", "url": "https://www.e-flux.com/journal/161/6776851/the-university-within-the-limits-of-automated-knowledge" } ], "stages": [ "higher" ], "summary": { "en": "Yuk Hui asks what universities lose when knowledge production and economic utility become their defining purposes.", "zh": "许煜追问：若大学只按知识产出和经济效益定义自己，会失去什么？" }, "title": { "en": "The University Within the Limits of Automated Knowledge", "zh": "知识可以自动生产后，大学还为什么存在？" }, "topics": [ "theory", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Examples cover standards, reassessment variants, exemplars, practice and interpreting feedback.", "One feedback-decoding prompt produced a complete corrected proof. The metacognitive narrative example is hypothetical; his objection to automated grading is relational as well as practical." ], "zh": [ "教师端包括梳理标准、生成重试题和示例；学生端包括练习、理解反馈与组织学习反思。", "一次解释反馈的尝试直接给出了修正答案，暴露了支持容易变成代做的问题；反思叙事用途则明确只是设想。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "talbert-six-ai-grading-uses-20260302", "limits": { "en": "Personal trials and proposals, not learning-effect evidence; examples were not run and privacy claims were not legally validated.", "zh": "个人试用与建议，非学习效果证据；未运行示例程序，也未把其隐私判断当法律结论。" }, "publishedAt": "2026-03-02", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Robert Talbert · Grading for Growth", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://gradingforgrowth.com/p/six-ways-to-use-artificial-intelligence" } ], "stages": [ "higher" ], "summary": { "en": "Talbert separates AI assistance with design and practice from delegating grading itself.", "zh": "Talbert 区分帮助设计、练习和理解反馈，与把评分关系交给 AI。" }, "title": { "en": "Six ways to use artificial intelligence tools in alternative grading", "zh": "替代评分中的六种 AI 用法及一道边界" }, "topics": [ "assessment", "teaching", "tools" ], "type": "practice" }, { "body": { "en": [ "Li discusses combining video with written expression and designing interactions among teachers, students and AI. He connects questioning to understanding and independent judgment.", "The interview asks who defines educational purposes when technology enters schools. It also acknowledges that mechanisms of human–AI co-creation still need systematic investigation." ], "zh": [ "他以视频观看后再用文字表达、组织人际与人机对话等例子，讨论技术怎样进入完整的教学设计。提问、理解和独立判断应相互支持。", "这为学校提出一个价值问题：采购与设计 AI 时，育人目标由谁确定？访谈同时承认，人机共创的机制仍需系统研究。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "li-zhengtao-education-shapes-ai-20260222", "limits": { "en": "Educational philosophy and interview claims; statements about irreplaceability are not established capability bounds. English is editorial paraphrase.", "zh": "教育哲学与访谈主张；“无法替代”等判断并非经研究确立的能力上限。英文内容为策展转述。" }, "publishedAt": "2026-02-22", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "李政涛 / 谷晓丹 · 澎湃新闻", "scope": { "en": "Read the full ECNU repost; checked the original publisher for author, title and date. Video not watched.", "zh": "阅读全文的华东师大官方转载，并回溯澎湃原始页核对记者、标题与日期；未看视频。" }, "status": "verified", "url": "https://m.thepaper.cn/newsDetail_forward_32592743" }, { "checkedAt": "2026-10-07T09:39:29Z", "name": "华东师范大学教育学部 · 官方全文转载", "scope": { "en": "Read the complete published Q&A text.", "zh": "核读完整问答正文。" }, "status": "verified", "url": "https://ed.ecnu.edu.cn/6a/74/c45970a748148/page.htm" } ], "stages": [ "school", "higher" ], "summary": { "en": "Li Zhengtao argues that educational purposes should shape technology, protecting students’ own thinking and lived learning.", "zh": "李政涛在长访谈中主张：教育的目标应反过来影响技术设计，保留学生亲身思考与成长的过程。" }, "title": { "en": "Education in the AI Era Should Foster Independent Thought and Lived Growth", "zh": "李政涛：教育也要向 AI 提出要求" }, "topics": [ "theory", "teaching", "learning" ], "type": "opinion" }, { "body": { "en": [ "Required exit tickets feed a thematic analysis combined with course materials; he reports 15–20 minutes of preparation per session.", "He credits Craig Kapp and Rob Egan’s workshop, then describes his own adaptation and how responding visibly encourages participation." ], "zh": [ "他把约50名学生的必做离堂反馈导入机构工具，结合课程资料归纳主题并准备补充活动，报告每次约15至20分钟。", "文章明确方法来自 Craig Kapp 与 Rob Egan 的工作坊；其增量是自己的课堂工作流与调整。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "ipeirotis-exit-tickets-20260215", "limits": { "en": "Self-reported workflow efficiency, not measured learning gain; data-protection claims concern his NYU environment.", "zh": "教师效率自报，未测学习增益；数据保护说法仅属其NYU机构环境。" }, "publishedAt": "2026-02-15", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Panos Ipeirotis · A Computer Scientist in a Business School", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://www.behind-the-enemy-lines.com/2026/02/listening-to-my-students-at-scale-exit.html" } ], "stages": [ "higher" ], "summary": { "en": "Ipeirotis describes turning exit-ticket responses into targeted follow-up teaching.", "zh": "Ipeirotis 记录如何把学生困惑、理解与要点整理成下一课的材料。" }, "title": { "en": "Listening to My Students at Scale: Exit Tickets, NotebookLM, and the Tightest Feedback Loop I’ve Ever Built", "zh": "用离堂反馈及时调整下一堂课" }, "topics": [ "assessment", "teaching", "tools" ], "type": "practice" }, { "body": { "en": [ "Flexible pathways complicate program-wide judgments; aggregated grades can obscure capability.", "The proposal uses shared outcomes, backbone units, milestones and visible feedback records, adapted to each degree’s structure." ], "zh": [ "BA、BSc 等学位的课程组合与顺序各异，使简单汇总分数难以说明毕业能力。", "他们建议明确学位目标、确定骨干单元、串联反馈并用作品集或视图累积证据，同时承认不能直接照搬专业学位模式。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "bugeja-generalist-programme-assessment-20260215", "limits": { "en": "Conceptual and practical proposal without a reported implementation evaluation.", "zh": "概念与实践建议，未报告系统实施评估。" }, "publishedAt": "2026-02-15", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Hayley Bugeja, Hora Hedayati & Diana Turnip · Needed Now", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://needednowlt.substack.com/p/strengthening-validity-and-trust" } ], "stages": [ "higher" ], "summary": { "en": "The authors address fragmented evidence in flexible generalist degrees.", "zh": "作者聚焦自由选课与跨学科路径造成的证据断裂，提出骨干课程和里程碑。" }, "title": { "en": "Strengthening validity and trust in generalist degrees through program-level approaches to assessment: Possible or pipe dream?", "zh": "通识型学位怎样建立跨课程评价证据？" }, "topics": [ "assessment", "policy" ], "type": "practice" }, { "body": { "en": [ "Mills retains detection among several strategies while reporting more false flags and stressful meetings. She rejects treating scores as conclusive, uses process conversations and offers alternatives; she also acknowledges possible survey-response bias.", "Compare tool performance, response procedures and student experience separately, rather than turning one teacher’s experience into a general rule." ], "zh": [ "Mills 继续把检测用于多种策略之一，却报告误报和紧张会谈增加。她不把检测当作确定证据，结合写作过程讨论，并提供替代安排；她也承认学生问卷可能受到师生权力关系影响。", "与反检测文章对读时，应把工具性能、处置程序及师生体验拆开评估，避免把任何一位教师的经验提升为通用规则。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "mills-detection-reconsidered-20260207", "limits": { "en": "Individual observations; detector accuracy and fairness unverified.", "zh": "个人课堂观察；未验证检测准确率或公平性。" }, "publishedAt": "2026-02-07", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Anna Mills · Substack", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://annamills.substack.com/p/a-brief-update-why-im-still-using" } ], "stages": [ "higher" ], "summary": { "en": "A bounded pro-detection stance with acknowledged costs.", "zh": "保留一个有边界、也有代价的不同立场。" }, "title": { "en": "A brief update: Why I’m still using AI detection after all, alongside many other strategies", "zh": "支持有限检测的教师，也报告更多误报" }, "topics": [ "assessment", "teaching", "tools" ], "type": "practice" }, { "body": { "en": [ "Mills describes shorter, focused feedback that protects students’ wording within peer review and reflection. The published prompt still contains conflicting five- and six-sentence limits, so it needs adaptation and testing.", "The important test is whether students understand, challenge and use feedback, rather than whether a system generates an impressive response." ], "zh": [ "Mills 介绍 PAIRR 提示修订：缩短反馈、每次集中一点、保留学生自己的措辞，并把 AI 放在同伴反馈和反思之中。公开提示仍混有五句与六句的冲突限制，不能直接当成已验证模板。", "最值得迁移的是测试过程：学生是否真的读懂、质疑并使用反馈，而不是系统是否生成了一大段貌似专业的建议。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "mills-pairr-prompts-20260127", "limits": { "en": "Prompts not run; satisfaction is not learning evidence.", "zh": "未运行提示；学生满意度不是学习效果。" }, "publishedAt": "2026-01-27", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Anna Mills · Substack", "scope": { "en": "Read body and published prompt; linked appendices not all opened.", "zh": "已读正文及公开提示；未打开所有附录。" }, "status": "verified", "url": "https://annamills.substack.com/p/ai-feedback-customized-for-student" } ], "stages": [ "higher" ], "summary": { "en": "Prompt revision driven by feedback overload.", "zh": "由学生反馈推动的提示设计修订。" }, "title": { "en": "AI feedback customized for student writers: the updated PAIRR prompts", "zh": "PAIRR 更新：反馈太多，也可能让学生不再读" }, "topics": [ "teaching", "assessment", "tools" ], "type": "practice" }, { "body": { "en": [ "Wexler contrasts writing one’s own outline to ease working-memory demands with asking AI to decide the structure. She acknowledges that the literature often uses offloading and outsourcing interchangeably.", "Her proposed distinction redirects attention to explicit writing instruction: students need ways to organize ideas themselves, rather than treating a smoother finished product as the educational goal." ], "zh": [ "Wexler 借 Kirschner 的观点区分自己生成提纲以减轻工作记忆负担，与让 AI 决定文章结构。她同时承认相关文献常混用 offloading 与 outsourcing，自己的划分是有意提出的分析框架。", "她把问题引向明确的写作教学：学生若学会组织句子和提纲，就更可能获得工具辅助记忆的好处，而不是只追求更顺滑的成品。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "wexler-offloading-outsourcing-20260118", "limits": { "en": "Terminological and pedagogical argument, not a consensus taxonomy or longitudinal causal study.", "zh": "术语与教学论证，不是已达成共识的分类或纵向因果研究。" }, "publishedAt": "2026-01-18", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Natalie Wexler / Minding the Gap", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://nataliewexler.substack.com/p/is-relying-on-ai-cognitive-offloading" } ], "stages": [ "school", "higher", "general" ], "summary": { "en": "Outlines illustrate a distinction between external memory support and delegating the thinking itself.", "zh": "用提纲说明外部记忆支持与代替思考的区别，同时保留术语争议。" }, "title": { "en": "Is Relying on AI Cognitive “Offloading” or “Outsourcing”?", "zh": "把想法写下来，与把思考交出去" }, "topics": [ "learning", "theory", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Groups rotate through prepared coding questions with structured prompts and follow-ups. Two instructors and a teaching assistant divide the work.", "The authors disclose a failed extra-time accommodation arrangement and its revision, making workload and accessibility central to adoption." ], "zh": [ "每组五至六人，按预设问题、提示和追问轮流展示；两位教师与一位助教分担提问、评分和技术工作。", "论文特别记录额外时间安排的投诉与修改，说明可扩展性不能只按考试时长判断。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "barba-stegner-conversational-exam-20260115", "limits": { "en": "Single-course report without independent validity or fairness validation; “cannot be faked” is the authors’ strong claim.", "zh": "单课程实施报告，无独立效度或公平性验证；“无法伪装”是作者强主张。" }, "publishedAt": "2026-01-15", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Lorena A. Barba & Laura Stegner · arXiv / Computer", "scope": { "en": "Read v1 HTML body and implementation checklist; no tools run or external references audited.", "zh": "通读v1 HTML正文及实施清单，未逐项运行或复核外链。" }, "status": "verified", "url": "https://arxiv.org/abs/2601.10691" } ], "stages": [ "higher" ], "summary": { "en": "Barba and Stegner report small-group live coding for 58 students, including substantial implementation costs.", "zh": "Barba 与 Stegner 用小组现场编程考58人，并公开备题、人力与合理调整难题。" }, "title": { "en": "The Conversational Exam: A Scalable Assessment Design for the AI Era", "zh": "对话式考试：现场编程、追问与规模化的真实代价" }, "topics": [ "assessment", "teaching" ], "type": "practice" }, { "body": { "en": [ "Gopnik distinguishes early exploration from school-age skill development through apprenticeship. Her concern is that schooling can reward getting good at school rather than doing meaningful work.", "She frames AI as a cultural technology whose reliability students must learn to judge. Cowen challenges her account using reasoning models. Their disagreement is part of the interview’s educational value." ], "zh": [ "她区分幼儿的探索与学龄儿童的技能练习，强调示范、亲自尝试和反馈，并担心学生只练成了应付学校的本领。", "她把生成式 AI 看作新的文化工具，主张教学生判断它何时提供可靠信息。Cowen 以推理模型反驳她，留下一个值得继续追问的分歧。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "gopnik-childhood-learning-cultural-ai-20251217", "limits": { "en": "An openly contested interview, not a settled account of schooling or current AI capabilities. Recording preceded publication.", "zh": "这是有明确反驳的访谈观点，不是对所有学校或当代 AI 能力的定论；录制时间早于发布时间。" }, "publishedAt": "2025-12-17", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Alison Gopnik / Tyler Cowen · Conversations with Tyler", "scope": { "en": "Close-read learning, schooling and AI sections of the official transcript; no audio/video playback or full reading of diagnostic/family sections.", "zh": "细读官方转录的儿童学习、学校教育及 AI 争论段落；未播放音视频，未精读全部诊断与家庭话题。" }, "status": "verified", "url": "https://conversationswithtyler.com/episodes/alison-gopnik/" } ], "stages": [ "school", "family", "higher" ], "summary": { "en": "Gopnik connects exploration, apprenticeship and AI literacy in an interview that preserves disagreement about machine reasoning.", "zh": "Gopnik 将儿童探索、学徒式练习和 AI 素养放进同一场讨论；主持人也直接挑战她的 AI 观点。" }, "title": { "en": "Alison Gopnik on Childhood Learning, AI as a Cultural Technology, and Rethinking Nature vs. Nurture", "zh": "Gopnik：学会探索、练习与使用文化工具" }, "topics": [ "learning", "theory", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Most standards required one successful in-class demonstration alongside an asynchronous attempt. Long-form proof work remained outside class.", "Students still reported stress; anonymous surveys revealed heterogeneous attitudes and uses. The course illustrates a different compromise from moving everything into exams." ], "zh": [ "每项标准通常只给一次课外尝试，另一次成功需在课堂测验中证明；只有三次大测验及期末回到现场。", "学生仍觉得现场测验更紧张。他的匿名调查也呈现不同用途与态度，不能把学生统称为同一种 AI 使用者。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "clark-small-changes-ai-20251215", "limits": { "en": "Single-course self-report and survey evidence; the author acknowledges his existing prohibition is difficult to enforce.", "zh": "单课程教师复盘；调查属自报，作者承认原有禁用声明难以执行。" }, "publishedAt": "2025-12-15", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "David Clark · Grading for Growth", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://gradingforgrowth.com/p/small-changes-to-handle-ai" } ], "stages": [ "higher" ], "summary": { "en": "Clark combines flexible practice with selected in-class checks while retaining proof portfolios.", "zh": "Clark 在证明写作课中结合课外练习与少量现场验证，保留长篇证明作品集。" }, "title": { "en": "Small changes to handle AI", "zh": "不推倒课程：用少量改动应对 AI" }, "topics": [ "assessment", "teaching" ], "type": "practice" }, { "body": { "en": [ "Across models and benchmarks, expert personas produce no dependable general gain, and some mismatched or low-knowledge personas worsen performance.", "For AI literacy, distinguish a role that shapes perspective or a simulation from evidence of factual competence. The report explicitly leaves room for useful non-accuracy purposes of role prompting." ], "zh": [ "团队比较多个模型与不同身份提示，发现专家设定的收益不稳定；部分不匹配或低知识身份还会降低表现。", "教学上可区分两件事：角色设定有助于语气、视角或模拟，却不因此保证事实正确。这样的区分比让学生背诵万能提示模板更值得讨论。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "gail-expert-personas-20251207", "limits": { "en": "Tests factual-question accuracy, not the full value of roleplay, writing style or pedagogical dialogue.", "zh": "客观题准确率测试，不能据此否定角色模拟、写作语气或教学对话的所有用途。" }, "publishedAt": "2025-12-07", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Savir Basil, Ina Shapiro, Dan Shapiro, Ethan Mollick, Lilach Mollick, Lennart Meincke · Wharton GAIL", "scope": { "en": "Read the public technical report, including methods, exceptions and scope; not the SSRN attachment.", "zh": "阅读公开技术报告全文，包括模型、题库、例外结果与用途边界；未核读 SSRN 附件。" }, "status": "verified", "url": "https://gail.wharton.upenn.edu/research-and-insights/playing-pretend-expert-personas/" } ], "stages": [ "higher", "general" ], "summary": { "en": "This report tests whether assigning expert identities reliably improves answers to difficult factual questions.", "zh": "这项测试挑战常见提示技巧：给模型一个专家身份，并不稳定提升困难事实题的准确率。" }, "title": { "en": "Playing Pretend: Expert Personas Don’t Improve Factual Accuracy", "zh": "“扮演专家”能让答案更准确吗？" }, "topics": [ "teaching", "tools" ], "type": "research" }, { "body": { "en": [ "The agenda arose from a September 2025 workshop and connects questions about purposes, participants, evidence and settings.", "It asks researchers to describe concrete AI-mediated practices, build on earlier scholarship, and treat equity as a foundational concern." ], "zh": [ "作者从2025年9月研讨形成共同议程，要求区分为何评、评谁、评什么、怎样评，以及发生在哪里。", "他们特别反对把 AI 当单一变量，并提醒写作、反馈与分工研究的既有知识不应被“全新技术”叙事抹掉。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "corbin-assessment-research-agenda-20251203", "limits": { "en": "Conceptual agenda, not systematic review. Landing page says December 3; PDF says December 4.", "zh": "概念议程，非系统综述；网页日期12月3日，PDF印12月4日，两者差异保留。" }, "publishedAt": "2025-12-03", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Thomas Corbin and 15 co-authors · JUTLP", "scope": { "en": "Read journal page and PDF introduction, principles and agenda; references not audited.", "zh": "阅读期刊页及PDF的导言、原则和研究议程；未复核全部参考文献。" }, "status": "verified", "url": "https://open-publishing.org/journals/index.php/jutlp/article/view/2533" } ], "stages": [ "higher" ], "summary": { "en": "An assessment research agenda asks for precise practices, historical grounding and connected inquiry.", "zh": "研究团队提出以评价目的、具体 AI 用法和公平性组织研究，而非继续堆积孤立案例。" }, "title": { "en": "Assessment after Artificial Intelligence: The Research We Should Be Doing", "zh": "AI 之后的评价：我们应该研究哪些问题？" }, "topics": [ "assessment", "theory" ], "type": "opinion" }, { "body": { "en": [ "Christodoulou asks whether tutors elicit answers, generate accurate questions, improve on existing tutoring systems, and sustain the structure needed for students to persist.", "These are evaluation criteria, not proof that all AI tutoring fails. Her own question-generation difficulties illustrate the distance between fluent explanation and checking understanding." ], "zh": [ "作者要求 AI 导师证明：是否让学生实际答题；题目是否准确；比已有题库和智能辅导系统好在哪里；如何提供线下课堂的结构与坚持学习的条件。", "这是一套评估问题，而不是认定所有 AI 辅导必然无效。她用自己的出题失败说明，能写出漂亮解释与能检测学生理解之间有距离。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-tutor-four-tests-20251116", "limits": { "en": "Conceptual and implementation argument; comparisons such as textbook error rates are not measured here.", "zh": "认知与实施论证；关于教材错误率等比较是作者判断，不是本文测量。" }, "publishedAt": "2025-11-16", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/llm-tutors-are-they-any-good" } ], "stages": [ "school", "family" ], "summary": { "en": "Question quality, accuracy, alternatives and persistence matter beyond fluent explanations.", "zh": "问题质量、准确性、替代方案与坚持学习，比解释是否流畅更重要。" }, "title": { "en": "Can LLMs be personal tutors?", "zh": "AI 导师要回答的四个教学问题" }, "topics": [ "teaching", "learning", "theory" ], "type": "opinion" }, { "body": { "en": [ "Five UK schools compared static hints with interactive tutoring; tutoring sessions were randomized between humans and human-supervised LearnLM. Roughly 76% of drafted messages needed no substantive edits.", "The next-topic advantage was estimated at 5.5 percentage points, with a 95% credible interval of −1.4 to +12.4. This supports further study, not established autonomous superiority or long-term transfer." ], "zh": [ "五所英国中学把静态提示与交互辅导比较，辅导对话再随机由人类或受人类审核的 LearnLM 支持。约 76% 的 AI 草稿无需实质修改；测试的是协作系统。", "下一单元新题的成功率估计比纯人类辅导高 5.5 个百分点，但 95% 可信区间为 −1.4 至 +12.4。短期跨题表现值得追踪，不能称已证明长期迁移或自主 AI 胜过教师。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "learnlm-eedi-supervised-rct-20251111", "limits": { "en": "Vendor-involved exploratory study; short-term platform outcomes and message-level supervision, without long-term independent testing.", "zh": "厂商参与的探索性研究；短期平台指标、逐条人工监督，缺少长期独立测验。" }, "publishedAt": "2025-11-11", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Google LearnLM Team / Eedi", "scope": { "en": "Read official report pages 1–7: abstract, design, results, credible interval and limitations; not all appendices.", "zh": "已读官方报告第 1–7 页摘要、设计、结果、可信区间及局限；未核查全部附录。" }, "status": "verified", "url": "https://storage.googleapis.com/deepmind-media/LearnLM/learnLM_nov25.pdf" }, { "checkedAt": "2026-10-07T09:44:18Z", "name": "arXiv submission record", "scope": { "en": "Verified arXiv posting on 29 December 2025, later than the report’s 11 November cover date.", "zh": "核实同题论文 2025-12-29 上传，晚于报告封面的 2025-11-11。" }, "status": "verified", "url": "https://arxiv.org/abs/2512.23633" }, { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou: conditional response", "scope": { "en": "Read the 11 January 2026 response as evidence of changed author judgment, not a separate trial.", "zh": "读 2026-01-11 评论全文；作为作者改变判断的证据，不作独立试验。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/maybe-llm-tutors-might-be-able-to" } ], "stages": [ "school" ], "summary": { "en": "An exploratory 165-student trial retains message-by-message human supervision and uncertain transfer estimates.", "zh": "165 名学生的探索性试验保留逐条人类审核；迁移估计仍有较大不确定性。" }, "title": { "en": "AI tutoring can safely and effectively support students: An exploratory RCT in UK classrooms", "zh": "人类逐条审核的 AI 数学辅导：有希望，但不是自主导师证明" }, "topics": [ "teaching", "learning", "assessment" ], "type": "research" }, { "body": { "en": [ "No More Marking reports roughly 70,000 scripts from 177 secondary schools, with AI agreeing with 83% of 133,983 human comparisons. Most schools used approximately 90% AI judging.", "Pairwise comparison replaces direct marking while retaining teachers. Agreement on a jointly constructed scale still needs independent validation; reliability, bias, susceptibility to gaming and learning are separate questions." ], "zh": [ "No More Marking 报告约 7 万篇七至九年级作文、177 所学校的混合评价。教师作出 133,983 次比较判断，AI 与其一致的比例为 83%；多数学校使用约九成人工智能判断。", "它把单篇直接打分改成两篇比较，并保留教师参与。需要继续问：共同形成的分数尺度是否提供足够独立的验证？可靠性、偏差、可被操纵程度和学习收益仍是不同问题。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-comparative-judgment-20251025", "limits": { "en": "Vendor account, not independent validity evidence or a delayed, unaided writing test.", "zh": "厂商自述；并非独立效度研究，未验证延迟、无 AI 写作能力。" }, "publishedAt": "2025-10-25", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/ai-is-uncannily-good-at-judging-writing" }, { "checkedAt": "2026-10-07T09:44:18Z", "name": "No More Marking: author’s methodological clarification", "scope": { "en": "Read the author’s 9 November 2025 clarification that human and AI comparisons jointly construct the scale.", "zh": "读作者 2025-11-09 对分数尺度如何混合人机判断的说明。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/ai-is-uncannily-good-at-judging-writing/comments" } ], "stages": [ "school" ], "summary": { "en": "A large vendor report concerns judging agreement, not student learning.", "zh": "公司报告大规模人机判断一致性；一致不自动等于评价有效或学生学会。" }, "title": { "en": "AI is uncannily good at judging writing", "zh": "AI 比较判断的高一致率，能说明什么？" }, "topics": [ "assessment", "tools" ], "type": "vendor" }, { "body": { "en": [ "Examples show how exact matching, attractive presentation and averages can distort judgments of capability. Evaluation begins with deciding what the task actually requires.", "The slides recommend validating model judgments on a small scale before expanding use. They offer material for teaching how to scrutinize AI evidence, rather than evidence that AI improves student learning." ], "zh": [ "课件用具体例子展示：措辞不同可能被判错，漂亮表达可能赢得人类偏好，平均表现也可能遮住严重失误。选择评量方式，先要知道自己在意什么。", "它建议先小规模核对模型评审与人的判断，再考虑扩大使用。教师可把它作为理解 AI 证据的教学资源，而非学生学习效果的研究。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "hung-yi-lee-ai-evaluation-20251020", "limits": { "en": "Date is the video’s publication, distinct from the October 17 class. Slides are not a full transcript; not every figure or cited study was checked.", "zh": "发布日为视频上传日，课堂日期为2025-10-17。课件并非完整口述转录；未逐页核验全部图表或所引研究。" }, "publishedAt": "2025-10-20", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "李宏毅 · 国立台湾大学公开课程 / Hung-yi Lee", "scope": { "en": "Verified original-channel video metadata. Content summary uses official slide sections on evaluation, bias, prompting and leakage; video not played.", "zh": "核对原频道视频元数据；主要阅读依据是原站课件的评量、偏见、提示与泄漏段落。未播放视频。" }, "status": "limited", "url": "https://www.youtube.com/watch?v=dWQVY_h0YXU" }, { "checkedAt": "2026-10-07T09:39:29Z", "name": "李宏毅 · 第4讲官方课件（96页）", "scope": { "en": "Read extracted text on metrics, LLM judges, prompts and contamination; not all visual content verified.", "zh": "阅读文字提取中的评量、LLM裁判、提示与题库泄漏部分；部分视觉内容未核验。" }, "status": "limited", "url": "https://speech.ee.ntu.edu.tw/~hylee/GenAI-ML/2025-fall-course-data/Evaluation.pdf" }, { "checkedAt": "2026-10-07T09:39:29Z", "name": "国立台湾大学 · 2025秋课程目录", "scope": { "en": "Checked objectives, FAQ, schedule and official video/slide links.", "zh": "核对课程目标、FAQ、讲次与视频/课件的官方关联。" }, "status": "verified", "url": "https://speech.ee.ntu.edu.tw/~hylee/GenAI-ML/2025-fall.php" } ], "stages": [ "higher", "general" ], "summary": { "en": "This public lesson extends AI literacy to evaluation: metrics, human preferences, model judges and benchmark contamination.", "zh": "这堂公开课把 AI 素养延伸到评量：标准答案、人的偏好、模型裁判和题库泄漏，都会影响结果。" }, "title": { "en": "Pitfalls in Evaluating Generative AI — Hung-yi Lee, Lecture 4", "zh": "李宏毅：评量 AI 时，分数到底代表什么？" }, "topics": [ "teaching", "tools", "assessment" ], "type": "practice" }, { "body": { "en": [ "Invigilation can strengthen authorship evidence without proving broader understanding or transfer. Essays likewise require contextual evaluation.", "Steel proposes examining the assumptions behind each intended interpretation and revising them as evidence changes." ], "zh": [ "监考可能支持本人作答的判断，却未必代表学生最佳表现或迁移能力；论文也需结合具体设计、关系和其他证据判断。", "他建议明确每一种分数解释背后的假设，持续检查，而不是给某类任务贴上永久有效或无效标签。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "steel-plausible-validity-20251019", "limits": { "en": "Conceptual explanation, not validation of any particular assessment.", "zh": "概念性解释，不是任何特定任务已通过效度验证的证明。" }, "publishedAt": "2025-10-19", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Alex Steel · Needed Now", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://needednowlt.substack.com/p/valid-assessment-is-a-plausible-argument" } ], "stages": [ "higher" ], "summary": { "en": "Steel explains why assessment security alone does not establish a valid learning inference.", "zh": "Alex Steel 解释为何“安全”与“能证明学习”不是同一个判断。" }, "title": { "en": "Valid assessment is a plausible argument not an absolute", "zh": "有效评价是一项可辩护的论证，不是绝对标签" }, "topics": [ "assessment", "theory" ], "type": "opinion" }, { "body": { "en": [ "Warner’s campus visits suggest shared values, supported faculty exploration and room for disagreement. He supports space for both promising AI experiments and teaching without it, while acknowledging selection bias among institutions inviting him.", "A faculty-development program can ask whether it rewards adoption itself or reasoned investigation against educational goals." ], "zh": [ "Warner 从校园访问中归纳共同价值、受支持的教师探索及容纳分歧的重要性。他认为有意义的 AI 实验和不用 AI 的教学都应获得空间，同时主动承认邀请他的机构可能存在选择偏差。", "可用于审视教师发展计划：它是在要求人人采用一种技术，还是让不同学科用可公开讨论的目标来判断方案？" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "warner-progress-institutions-20251003", "limits": { "en": "Visit impressions, not a representative survey or outcome evaluation.", "zh": "访问印象，非代表性机构调查或效果评价。" }, "publishedAt": "2025-10-03", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "John Warner · Inside Higher Ed / Just Visiting", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://www.insidehighered.com/opinion/columns/just-visiting/2025/10/03/we-dont-need-retreat-challenge-ai-schools" } ], "stages": [ "higher" ], "summary": { "en": "A skeptical author identifies room for collective experimentation.", "zh": "强烈怀疑 AI 的作者，也记录可以合作的路径。" }, "title": { "en": "Making Progress on Teaching in a World With AI", "zh": "面对 AI，机构如何支持分歧中的共同探索？" }, "topics": [ "teaching", "policy" ], "type": "opinion" }, { "body": { "en": [ "Securing individual tasks does not automatically establish program outcomes without alignment.", "The guidance warns that a return to familiar timed exams can narrow authenticity and equity, and asks institutions to test their security assumptions." ], "zh": [ "报告说明三条路径都需要可靠证据，但把每门课“锁住”并不会自动保证学位目标。", "它也警告过度回归限时监考会缩减真实性和多样性；学校需检验所谓安全任务是否真的可靠。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "teqsa-enacting-reform-20250924", "limits": { "en": "Expert guidance, not outcome evaluation. The PDF’s 2026-06 storage path is not its September 2025 publication date.", "zh": "专家协作指引，非实施效果试验；当前PDF路径含2026-06，封面与发布页仍为2025年9月。" }, "publishedAt": "2025-09-24", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "TEQSA · Jason Lodge and collaborating contributors", "scope": { "en": "Read landing page and PDF pathways, challenges and critical questions; not every case audited.", "zh": "阅读发布页及PDF三路径、挑战和关键问题部分；未复核所有案例。" }, "status": "verified", "url": "https://www.teqsa.gov.au/guides-resources/resources/corporate-publications/enacting-assessment-reform-time-artificial-intelligence" } ], "stages": [ "higher" ], "summary": { "en": "TEQSA compares program-wide, unit-level and hybrid approaches to assurance of learning.", "zh": "TEQSA 比较整项学位、单门课程和混合路径，明确各自资源与公平性代价。" }, "title": { "en": "Enacting assessment reform in a time of artificial intelligence", "zh": "把评价改革落地：三条学习保证路径" }, "topics": [ "assessment", "policy" ], "type": "policy" }, { "body": { "en": [ "GeoAI connects AI-use boundaries to disciplinary goals: help with statistical concepts does not mean outsourcing climate-data interpretation. It suggests staged assignments, reflection and checked practice, with privacy and copyright considerations." ], "zh": [ "GeoAI 以地学作业说明如何连接 AI 使用规则与学习目标：可借助工具理解统计方法，但数据解释与结论仍由学生完成。页面建议分阶段提交、反思与形成性练习，并提醒检查生成内容、隐私和版权条件。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "serc-student-ai-use-20250922", "limits": { "en": "Teaching-design advice and examples, not tested effects for every strategy. Generated questions and feedback still need review.", "zh": "教学设计建议与示例，不是所有策略都经过效果试验；生成题目与反馈仍需审查。" }, "publishedAt": "2025-09-22", "sources": [ { "checkedAt": "2026-10-07T07:32:44.361022+00:00", "name": "SERC, Carleton College / GeoAI", "scope": { "en": "Public text read; page explicitly lists initial publication on 22 September 2025. This note is an original summary with attribution.", "zh": "公开正文已读，页面标明首次发布日期2025年9月22日；本条仅原创概述与链接。" }, "status": "verified", "url": "https://serc.carleton.edu/geo_ai/student_ai.html" } ], "stages": [ "higher" ], "summary": { "en": "GeoAI connects AI-use boundaries to disciplinary goals: help with statistical concepts does not mean outsourcing climate-data interpretation. It suggests staged assignments, reflection and checked practice, with privacy and copyright considerations.", "zh": "GeoAI 以地学作业说明如何连接 AI 使用规则与学习目标：可借助工具理解统计方法，但数据解释与结论仍由学生完成。页面建议分阶段提交、反思与形成性练习，并提醒检查生成内容、隐私和版权条件。" }, "title": { "en": "Supporting Appropriate Student Use of AI", "zh": "GeoAI：先明确学习目标，再划定学生使用 AI 的范围" }, "topics": [ "teaching", "assessment", "learning" ], "type": "practice" }, { "body": { "en": [ "The report examines peer and AI feedback, reflection and revision for 34 business-writing students; 10 of 46 enrolled across two sections opted out. Examples show students challenging generic advice and protecting their voice.", "The evidence supports examining implementation and experience. Self-report, revision quality and later unaided transfer remain different questions." ], "zh": [ "论文报告商务写作中的 PAIRR：同一草稿接受同伴与 AI 反馈，学生比较、反思并修订。分析聚焦 34 名学生；两班共 46 人中有 10 人选择不参与。文中展示学生质疑笼统建议、维护自身声音的例子。", "这些材料说明可实施的过程和参与者体验。判断效果时，仍需区分自述、实际修改质量以及离开 AI 后能否迁移。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "pairr-business-writing-20250916", "limits": { "en": "No randomized control; 34-person analysis differs from cited 654-person project.", "zh": "无随机对照；34 人分析不等于另引的 654 人项目。" }, "publishedAt": "2025-09-16", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "MacArthur et al. · Frontiers in Communication / UC Davis PAIRR", "scope": { "en": "Read abstract, course, implementation, results, conclusion and relevant footnotes.", "zh": "已读摘要、课程、实施、结果、结论及相关脚注。" }, "status": "verified", "url": "https://www.frontiersin.org/journals/communication/articles/10.3389/fcomm.2025.1615752/full" } ], "stages": [ "higher" ], "summary": { "en": "A 34-student course report centered on feedback comparison.", "zh": "34 名参与者的课程报告，保留真人反馈与选择。" }, "title": { "en": "GIFT-AI: teaching the game and leveling the field", "zh": "PAIRR 商务写作案例：比较同伴与 AI，再决定如何改" }, "topics": [ "teaching", "assessment", "learning" ], "type": "research" }, { "body": { "en": [ "The authors analyse late-2024 interviews through a wicked-problems framework. They identify conflicting goals rather than a universal fix.", "Their practical proposal is institutional permission to compromise, diverge and iterate, with support for the work involved." ], "zh": [ "研究者以棘手问题框架分析2024年下半年的访谈，认为制度不应要求教师找到永久适用的完美方案。", "他们提出允许妥协、因情境不同而分化，以及持续迭代，且应把改制劳动纳入支持安排。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "corbin-wicked-assessment-20250903", "limits": { "en": "One Australian university, purposive recruitment and deductive analysis; not a reform effectiveness trial.", "zh": "单所澳大利亚大学、目的性招募与演绎分析；不是改革效果试验。" }, "publishedAt": "2025-09-03", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Thomas Corbin, Margaret Bearman, David Boud & Phillip Dawson", "scope": { "en": "Read repository PDF abstract, methods, results examples and discussion; references not individually checked.", "zh": "阅读机构存档PDF的摘要、方法、结果案例及讨论；未逐条复核参考文献。" }, "status": "verified", "url": "https://opus.lib.uts.edu.au/rest/bitstreams/980f72c9-50a8-4632-8a04-823a5858b791/retrieve" } ], "stages": [ "higher" ], "summary": { "en": "Interviews with 20 educators reveal competing assessment goals and unavoidable trade-offs.", "zh": "20 位教师访谈显示，可信评价、包容性、职业准备与工作量常相互牵制。" }, "title": { "en": "The wicked problem of AI and assessment", "zh": "AI 与评价为何是没有唯一解的棘手问题？" }, "topics": [ "assessment", "theory", "policy" ], "type": "research" }, { "body": { "en": [ "Christodoulou supports workplace AI use but rejects copying expert workflows as a default pedagogy. Experts’ quick checks draw on accumulated knowledge that novices have yet to develop.", "Her argument asks which independent capabilities must precede collaboration with AI. It challenges curricula organized around producing polished outputs rather than deliberately building expertise." ], "zh": [ "Christodoulou 支持专家在工作中使用 AI，却反对据此让学生主要练习生成和审阅成品。她认为专家的快速判断来自长期知识积累，初学者并不具备同样基础。", "文章借刻意练习与认知外包提出课程分歧：学校应先建设哪些独立能力，再安排人机协作？这与直接把职业工作流搬进课堂的主张形成清楚对照。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-learning-not-workflow-20250731", "limits": { "en": "Argument and prediction, not a comparative curriculum experiment; the author develops AI assessment products.", "zh": "理论论证与预测，不是直接比较两类课程的实验；作者经营 AI 评价产品。" }, "publishedAt": "2025-07-31", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/education-isnt-working" } ], "stages": [ "school", "higher", "general" ], "summary": { "en": "The essay distinguishes productive work from the practice needed to become an expert.", "zh": "作者区分完成工作与训练能力，反对让初学者只练习审阅 AI 成品。" }, "title": { "en": "Education isn’t working", "zh": "为 AI 工作做准备，不等于照搬专家工作流" }, "topics": [ "learning", "theory", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Mills recommends challenging feedback, requesting clarification, adding missing concerns and eventually seeking human readers. Her own example shows repeated chatbot praise reinforcing an over-detailed workshop description until a colleague intervened.", "Use the piece to practice explaining acceptance and rejection of advice. More feedback does not automatically mean better revision." ], "zh": [ "Mills 建议学生追问、反驳、补充遗漏，再暂停并寻求真人意见。她讲述自己被 Claude 的连续赞许带偏，写出过度详细的活动介绍，最后由真人合作者纠正的经历。", "可把这当成反馈素养练习：学生需要说明采纳或拒绝建议的理由。收到更多建议，并不等于作出了更好的修改。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "mills-using-ai-feedback-20250719", "limits": { "en": "Guidance and self-report; course-specific rules still apply.", "zh": "教学指导与自述；使用前仍需遵守具体课程规则。" }, "publishedAt": "2025-07-19", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Anna Mills · Substack", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://annamills.substack.com/p/getting-the-most-out-of-ai-feedback" } ], "stages": [ "higher" ], "summary": { "en": "Student-facing strategies, including a failure of flattery.", "zh": "给学生的反馈对话方法，也包含作者自己的失误。" }, "title": { "en": "Getting the most out of AI feedback", "zh": "如何对 AI 反馈说不，并回到真人读者" }, "topics": [ "teaching", "learning", "tools" ], "type": "practice" }, { "body": { "en": [ "Unspecified tasks default to permitted-use category 2, with a recommendation to check with the tutor.", "The guidance covers acknowledgment, reasonable adjustments, data protection and reflection on whether AI displaced necessary learning." ], "zh": [ "未标类别时可先按第二类理解，但建议向导师核实；课程具体说明仍重要。", "指引要求说明生成输出的使用方式，同时讨论合理调整、保护资料和反思是否跳过了关键学习。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "ou-student-ai-guidance-20250715", "limits": { "en": "Date is the stated revision date, not initial release. These rules are institution-specific.", "zh": "日期是明确修订日，不是首次发布；OU规则不能推广为其他学校规则。" }, "publishedAt": "2025-07-15", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "The Open University", "scope": { "en": "Read student guidance, categories, disclosure and reasonable-adjustment sections.", "zh": "阅读学生指引正文、三分类及披露和调整部分。" }, "status": "verified", "url": "https://about.open.ac.uk/policies-and-reports/policies-and-statements/generative-ai-learning-teaching-and-assessment-ou-0" } ], "stages": [ "higher" ], "summary": { "en": "OU distinguishes prohibited, permitted and required use, with disclosure and accessibility guidance.", "zh": "OU 用三类要求区分禁止、允许与必须使用，并加入成人远程学习的执行细节。" }, "title": { "en": "Generative AI for students", "zh": "开放大学学生 AI 指引：允许范围、披露与合理调整" }, "topics": [ "assessment", "policy" ], "type": "policy" }, { "body": { "en": [ "Willingham distinguishes introductory knowledge goals from advanced argument-building. He describes reading essays during individual conferences, asking students to explain choices and giving immediate feedback.", "Conferences expose reasoning and permit emotional support; students reportedly find them stressful and valuable. His university experience does not establish comparable costs or outcomes in large school classes." ], "zh": [ "Willingham 区分初阶课程记忆概念与高阶课程整合论证的目标。他主张学生面谈解释、捍卫写作选择，并描述自己在面谈中首次阅读文章、即时反馈的做法。", "这种安排既让思考可见，也能回应情绪与动机；作者说学生感到紧张又有帮助。它是本人教学经验，不能直接推定中小学大班实施成本或效果相同。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "willingham-conferencing-20250708", "limits": { "en": "University teaching account and mechanism argument; no trial of conferencing effectiveness.", "zh": "大学教师的自述和机制论证；没有面谈效果试验。" }, "publishedAt": "2025-07-08", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daniel Willingham", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://dtwuva.substack.com/p/how-chatgpt-might-inadvertently-improve" } ], "stages": [ "higher", "general" ], "summary": { "en": "Willingham describes conferencing as a way to retain and inspect writing-related thinking.", "zh": "Willingham 用自己的大学课程说明，追问与反馈可以保留写作的认知工作。" }, "title": { "en": "How ChatGPT Might Inadvertently Improve Student Thinking", "zh": "面谈如何让写作背后的思考可见" }, "topics": [ "teaching", "assessment", "learning" ], "type": "practice" }, { "body": { "en": [ "The essay distinguishes answer-seeking from pedagogically structured support and acknowledges that added instructional support complicates comparisons. It also recommends generating one’s own ideas before consulting AI.", "A useful discussion prompt is to separate immediate performance, independent learning and diversity of ideas. These are different outcomes, rather than one blanket verdict on thinking." ], "zh": [ "文章对比直接索取答案与有教学设计的辅导，并提醒研究中的额外教学支持可能影响结果。它也提出先形成自己的想法，再引入 AI，减少被首轮建议限定。", "用于教师研讨时，可以分别检查任务表现、独立学习与创意多样性，而不把这些都塞进“思考变好或变坏”的单一判断。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "mollick-against-brain-damage-20250707", "limits": { "en": "Research commentary and practice suggestions, not a medical conclusion or settled long-term cognitive evidence.", "zh": "研究评论与实践建议，不是医疗判断；不能据此断言长期使用的所有认知影响已有定论。" }, "publishedAt": "2025-07-07", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Ethan Mollick · One Useful Thing", "scope": { "en": "Read the full essay, including learning, creative and collective-thinking sections; not all linked studies.", "zh": "阅读全文，包括学习、创造及集体思考三部分；未核读全部链接研究。" }, "status": "verified", "url": "https://www.oneusefulthing.org/p/against-brain-damage" } ], "stages": [ "general", "higher", "school" ], "summary": { "en": "Mollick challenges sensational interpretations of AI studies and examines learning, idea generation and writing practices.", "zh": "Mollick 反对把个别研究概括成“AI 伤脑”，转而讨论怎样保留学习、构思与写作中的思考。" }, "title": { "en": "Against “Brain Damage”", "zh": "讨论 AI 与思考，先问使用过程" }, "topics": [ "learning", "teaching", "theory" ], "type": "opinion" }, { "body": { "en": [ "Classes at one Turkish high school received a basic GPT-4 interface, teacher-designed tutoring safeguards, or no AI. Both AI versions improved practice scores; the basic version reduced subsequent unaided scores by about 17% relative to control.", "Safeguards largely removed that harm but did not demonstrate an unaided advantage. Tool-assisted success and learning must remain distinct outcomes." ], "zh": [ "土耳其一所高中的班级随机分配到普通 GPT-4 界面、教师设计护栏的导师或无 AI 组。练习阶段两种 AI 都提高分数；随后无工具测验中，普通 AI 组比对照低约 17%。", "有护栏导师基本消除了负面差异，但未显示无辅助测验的显著正收益。结果支持检验具体教学设计，不能概括成所有 AI 必然伤害学习或所有护栏都提高学习。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "bastani-guardrails-learning-20250625", "limits": { "en": "One school, four review sessions and immediate similar-problem exams; not long-term retention or far transfer.", "zh": "一校、四次课内复习；随后相似题考试，不是长期保持或远迁移证据。" }, "publishedAt": "2025-06-25", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Bastani et al. / PNAS", "scope": { "en": "Read abstract, experimental design, main results and Table 1; no code rerun or exhaustive supplement audit.", "zh": "已读摘要、实验设计、主要结果和表 1；未复跑代码或逐项审核补充材料。" }, "status": "verified", "url": "https://www.pnas.org/doi/10.1073/pnas.2422633122" } ], "stages": [ "school" ], "summary": { "en": "A mathematics RCT directly separates assisted performance from subsequent unaided assessment.", "zh": "高中数学随机试验直接区分辅助表现与随后无辅助测验。" }, "title": { "en": "Generative AI without guardrails can harm learning: Evidence from high school mathematics", "zh": "做题有 AI 更好，撤掉 AI 后却可能更差" }, "topics": [ "learning", "assessment", "theory" ], "type": "research" }, { "body": { "en": [ "He proposes unbundling exams by learning target, protecting reassessment time, and modelling deliberate practice.", "This is a provisional response, with a useful unresolved question: does increasing testing frequency offset the benefits of the grading method?" ], "zh": [ "建议包括按清晰标准拆分考试、预留补测时间，以及示范如何练习，而不只让学生反复再试。", "他把限时考试称作过渡办法，承认测试频率本身可能抵消替代评分原有的减压作用。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "talbert-test-forward-20250602", "limits": { "en": "Practice reflection, not treatment guidance or an effectiveness trial. Reader comments are not independent validation.", "zh": "实践反思，不是焦虑治疗建议或效果试验；评论区经验也不构成独立验证。" }, "publishedAt": "2025-06-02", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Robert Talbert · Grading for Growth", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://gradingforgrowth.com/p/alternative-grading-in-a-test-forward" } ], "stages": [ "higher" ], "summary": { "en": "Talbert examines reassessment backlogs, anxiety and practice in a test-heavy course.", "zh": "Talbert 拆解重考积压、考试焦虑与机械重复的实际问题。" }, "title": { "en": "Alternative grading in a test-forward environment", "zh": "限时考试回归后，替代评分怎样继续支持学习？" }, "topics": [ "assessment", "teaching" ], "type": "practice" }, { "body": { "en": [ "Open tasks support learning and feedback; secured tasks provide capability evidence. The latter need not exclude every AI use.", "The authors answer criticism of a binary model and frame the approach around trust and agency." ], "zh": [ "开放轨用于学习和反馈，允许学生选择 AI 用法；验证轨用于可信地判断能力，也不必等同于完全无 AI。", "作者回应“过于二元”的批评，但对信任和毕业能力的主张仍需在实施中检验。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "liu-bridgeman-trust-agency-20250601", "limits": { "en": "Designers’ explanation, not independent evaluation; grounded in Sydney’s 2025 context.", "zh": "制度设计者的解释，非独立成效评估；适用于文中2025年悉尼制度情境。" }, "publishedAt": "2025-06-01", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Danny Liu & Adam Bridgeman · Needed Now", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://needednowlt.substack.com/p/why-banning-ai-in-assessment-is-all" } ], "stages": [ "higher" ], "summary": { "en": "Sydney’s designers explain the rationale for open learning tasks alongside secured capability checks.", "zh": "两位制度设计者解释开放学习与受监督能力验证如何分工。" }, "title": { "en": "Why banning the banning of AI in assessment is all about increasing trust and agency", "zh": "悉尼大学两轨评价：为什么开放作业不能简单禁 AI？" }, "topics": [ "assessment", "policy" ], "type": "practice" }, { "body": { "en": [ "The panel contrasts open exploration, independent French drafting followed by AI editing, and interactive review. Jirout reports that vague prompts initially worked unevenly and were revised with students; Simotas acknowledges unknown long-term consequences.", "Boundaries should follow the learning task. Open assignments still need preparation, and enthusiasm or improved artifacts should not be mistaken for durable learning." ], "zh": [ "访谈对照开放式探索、外语课先独立起草再比较 AI 修改，以及教育学中的互动复习。Jirout 说早期模糊提示效果不一，后来与学生共同修订；Simotas 明确承认长期学习影响仍不清楚。", "可借鉴的是按任务目标拆分流程，而非复制统一口号。开放任务也需要先前训练；课堂兴趣与成品改进应和能力增长分开。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "bruff-uva-ai-assignments-20250520", "limits": { "en": "Four faculty accounts without a shared controlled design.", "zh": "四位教师自述；没有共同对照设计。" }, "publishedAt": "2025-05-20", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Derek Bruff · Intentional Teaching", "scope": { "en": "Read assignment and risk-discussion transcript sections.", "zh": "已读公开转录的作业与风险讨论段落。" }, "status": "verified", "url": "https://intentionalteaching.buzzsprout.com/2069949/episodes/17190682-ai-integrated-assignments-with-kiera-allison-jamie-jirout-spyros-simotas-jun-wang" } ], "stages": [ "higher" ], "summary": { "en": "Different disciplines, different boundaries and scaffolds.", "zh": "不同学科的开放探索、先独立后辅助与迭代提示。" }, "title": { "en": "AI-Integrated Assignments with Kiera Allison, Jamie Jirout, Spyros Simotas, & Jun Wang", "zh": "四位教师如何给 AI 作业安排边界" }, "topics": [ "teaching", "assessment" ], "type": "practice" }, { "body": { "en": [ "Volunteers in nine schools were randomized to a teacher-supported Copilot programme or ordinary schooling. The report estimates 0.23 SD in English and 0.31 SD on a composite, measured by supervised paper tests.", "Of 1,328 randomized students, 759 completed the final assessment; authors examine attrition sensitivity. The intervention bundles AI, instruction, practice and infrastructure, preventing attribution to chatbot access alone." ], "zh": [ "九所学校的志愿学生被随机分配到课后 Copilot 项目或常规教学。报告显示英语提高 0.23 标准差，含 AI 与数字知识的综合指标提高 0.31；测验为监考纸笔选择题。", "最初 1,328 人随机分组，759 人完成终测，作者做了流失敏感性分析。收益来自教师、额外练习、设备和 AI 的组合；不能说单独开放聊天机器人就能复制结果。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nigeria-copilot-rct-20250519", "limits": { "en": "Working paper; short-term outcomes, volunteer participation and substantial attrition. Schooling-equivalent years are a conversion, not observed retention.", "zh": "工作论文；短期结果、志愿参与及显著流失。所谓学习年数是换算，不是观察多年后的保持。" }, "publishedAt": "2025-05-19", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "De Simone et al. / World Bank", "scope": { "en": "Downloaded official PDF; read abstract, sampling and measurement, results, attrition analyses and discussion selectively; no replication.", "zh": "已下载并抽读官方 PDF 摘要、2.2–2.3 样本与测量、结果和流失敏感性、讨论；未复跑。" }, "status": "verified", "url": "https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099548105192529324" }, { "checkedAt": "2026-10-07T09:44:18Z", "name": "World Bank official report PDF", "scope": { "en": "Official PDF downloaded successfully and selectively read; web viewer failed on file size.", "zh": "官方 PDF 成功下载，正文抽读；网页工具因文件过大失败。" }, "status": "verified", "url": "https://documents1.worldbank.org/curated/en/099548105192529324/pdf/IDU-c09f40d8-9ff8-42dc-b315-591157499be7.pdf" }, { "checkedAt": "2026-10-07T09:44:18Z", "name": "World Bank author publication list", "scope": { "en": "Official author publication list gives report date as 19 May 2025.", "zh": "作者官方作品列表明确列报告日期 2025-05-19。" }, "status": "verified", "url": "https://www.worldbank.org/en/about/people/m/maria-barron" } ], "stages": [ "school" ], "summary": { "en": "A teacher-supported six-week programme improved paper-test outcomes, with important attrition and design caveats.", "zh": "教师支持的六周项目在纸笔测验上改善表现，但流失与组合干预需保留。" }, "title": { "en": "From Chalkboards to Chatbots: Evaluating the Impact of Generative AI on Learning Outcomes in Nigeria", "zh": "尼日利亚课后辅导：积极结果属于整套教学方案" }, "topics": [ "learning", "teaching" ], "type": "research" }, { "body": { "en": [ "The paper argues that increasingly detailed AI-use instructions cannot by themselves establish compliance or capability.", "It motivates changes to task mechanics. Later work on wearable AI and contextual validity should be read alongside its proposal." ], "zh": [ "他们认为更细的允许使用规则仍无法单独证明学生遵守了规则，因此评价证据需要来自任务机制。", "这是支持结构性改革的概念论证；后来关于可穿戴 AI 与效度论证的文章进一步暴露其边界。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "corbin-talk-is-cheap-20250515", "limits": { "en": "Read publisher-indexed abstract and substantial argument excerpts; direct fetch returned 403. Not an effectiveness demonstration.", "zh": "本轮仅读出版社索引的摘要和主要论证段落；直取403。不是结构性方案效果证明。" }, "publishedAt": "2025-05-15", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Thomas Corbin, Phillip Dawson & Danny Liu", "scope": { "en": "Publisher-indexed main argument excerpts; direct full text unavailable.", "zh": "出版社搜索索引中的主要论证段落；未取得直取全文。" }, "status": "limited", "url": "https://www.tandfonline.com/doi/abs/10.1080/02602938.2025.2503964" } ], "stages": [ "higher" ], "summary": { "en": "The authors distinguish compliance-dependent instructions from changes to assessment structure.", "zh": "Corbin、Dawson 与 Liu 区分依赖自愿遵守的说明，与改变任务本身的结构性设计。" }, "title": { "en": "Talk is cheap: why structural assessment changes are needed for a time of GenAI", "zh": "只改 AI 使用说明，为什么不足以保证评价效度？" }, "topics": [ "assessment", "theory", "policy" ], "type": "opinion" }, { "body": { "en": [ "Students combined generated answers with repeated autograder submissions. Demonstrations and verbal guidance did not produce the engagement she expected.", "Her redesign moved graded practice into class, retaining AI access under observation and peer interaction." ], "zh": [ "她观察到学生把生成答案与反复试提交结合，出勤和课堂关系恶化；原本的示范与口头提醒没有达到预期。", "后续改为现场小测与合作练习，让工具使用与教师观察并存，仍把如何平衡帮助与独立思考视为未解难题。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "barba-embracing-genai-failure-20250503", "limits": { "en": "Instructor account without randomised controls. Date is public author-manuscript deposit, not draft month or journal issue.", "zh": "教师自述，没有随机对照；日期为作者稿公开存档日，非稿上“2025年3月”或8月期号。" }, "publishedAt": "2025-05-03", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Lorena A. Barba · author manuscript / IEEE Computer", "scope": { "en": "Read author manuscript body and prompt appendix; checked public Figshare metadata.", "zh": "阅读作者存档PDF正文与系统提示附录，并查Figshare公开元数据。" }, "status": "verified", "url": "https://figshare.com/articles/journal_contribution/Experience_embracing_genAI_in_an_engineering_computations_course_What_went_wrong_and_what_next/28926647" } ], "stages": [ "higher" ], "summary": { "en": "Barba candidly describes a course chatbot and autograder enabling answer-seeking shortcuts.", "zh": "Barba 公开复盘课程聊天机器人与自动评分结合后出现的绕过学习现象。" }, "title": { "en": "Experience embracing genAI in an engineering computations course: What went wrong and what next", "zh": "拥抱 AI 的工程计算课：哪里出了问题？" }, "topics": [ "assessment", "teaching", "learning" ], "type": "practice" }, { "body": { "en": [ "The study includes 654 participants and qualitative analysis of 131 students’ reflections. Most preference respondents favored combined feedback, while many preferred peers alone. The authors stress that AI’s perceived utility occurred alongside human feedback.", "Perceptions, survey attrition and differing course support limit interpretation. The later business-writing report is related evidence, not an independent replication." ], "zh": [ "研究涉及 654 名参与者，并对 131 人的反思作定性分析。多数反馈偏好回答者选择同伴与 AI 结合，也有不少人只偏好同伴。作者强调，AI 的有用性是在保留真人反馈的情境中出现的。", "研究以感知和反思为主，存在问卷流失与课程支持差异，未直接评价反馈偏差。它不能证明写作能力提升，也不是后来商务写作案例的独立重复验证。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "pairr-formative-assessment-20250423", "limits": { "en": "No direct writing-gain or AI-feedback bias evaluation.", "zh": "未直接测量写作增益或评估 AI 反馈偏差。" }, "publishedAt": "2025-04-23", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Sperber et al. · Computers and Composition / UC eScholarship", "scope": { "en": "Read public PDF abstract, methods, main results and limitations.", "zh": "已读公开 PDF 的摘要、方法、主要结果与局限。" }, "status": "verified", "url": "https://escholarship.org/content/qt9v91m137/qt9v91m137.pdf" } ], "stages": [ "higher" ], "summary": { "en": "A mixed-methods study of experience and judgment.", "zh": "多课程混合方法研究，重点是体验和判断过程。" }, "title": { "en": "Peer and AI Review + Reflection (PAIRR): A human-centered approach to formative assessment", "zh": "PAIRR 原始研究：AI 反馈为何仍需同伴与反思？" }, "topics": [ "assessment", "teaching", "learning" ], "type": "research" }, { "body": { "en": [ "Wexler questions a university-writing study’s interpretation: better submitted prose neither establishes independent skill nor verifies ChatGPT use. She also challenges permanent simplification of reading material.", "She allows roles for AI in supporting educators but demands learning-relevant measures. The useful distinction is among temporary access support, permanently lowered text demands, and students’ own intellectual practice." ], "zh": [ "文章针对一项大学写作研究的解读提出批评：提交作品更好，并不能说明学生独立能力增强，也不能证明其确实用了 ChatGPT。她还质疑把简化阅读材料当作长期解决方案。", "她承认 AI 可以帮助教师和辅导员，但要求说明学习机制与结果指标。用于学校讨论时，应区分临时可及性支持、长期文本降级和学生自己的思考练习。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "wexler-output-not-learning-20250420", "limits": { "en": "Commentary; the criticized paper was not fully reviewed here and individual AI use cannot be inferred.", "zh": "评论文章；所评原论文未在本轮完整审读，不能凭本文判定所有学生如何使用 AI。" }, "publishedAt": "2025-04-20", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Natalie Wexler / Minding the Gap", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://nataliewexler.substack.com/p/do-students-who-rely-on-chatgpt-benefit" } ], "stages": [ "higher", "school" ], "summary": { "en": "Wexler challenges interpreting better submitted prose as improved writing ability.", "zh": "Wexler 质疑把提交文本质量提升直接解释为学生学会写作。" }, "title": { "en": "Do Students Who Rely on ChatGPT “Benefit”?", "zh": "作文变好了，学生是否真的受益？" }, "topics": [ "learning", "assessment", "theory" ], "type": "opinion" }, { "body": { "en": [ "Timed testing reduced his uncertainty about submitted work, but reattempts did not remove severe anxiety for some students.", "He questions excessive reassessment opportunities and the loss of higher-order work, arguing that productive practice needs explicit teaching." ], "zh": [ "课堂考试让他不再反复猜测作业作者，但即使能重考，部分学生仍有强烈焦虑。", "他认为频繁重试有时使学生跳过反馈；更深的问题是怎样保留高阶任务，并明确教会学生有效练习。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "talbert-ai-grading-reflection-20250407", "limits": { "en": "No controlled comparison; observations do not establish reduced cheating or causal anxiety effects.", "zh": "非对照研究；教师观察不能量化焦虑变化或证明作弊减少。" }, "publishedAt": "2025-04-07", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Robert Talbert · Grading for Growth", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://gradingforgrowth.com/p/my-ai-driven-grading-changes-a-3x3x3" } ], "stages": [ "higher" ], "summary": { "en": "A follow-up finds practical relief but persistent test anxiety and narrower assessment.", "zh": "三个月后，Talbert 发现安心与较轻批改负担，并没有消除考试焦虑。" }, "title": { "en": "My AI-driven grading changes: A 3x3x3 reflection", "zh": "改用考试之后：AI 驱动评分改制的复盘" }, "topics": [ "assessment", "teaching" ], "type": "practice" }, { "body": { "en": [ "Warner argues that peer response teaches through giving feedback, examining another draft and reconsidering one’s own choices. He accepts that some writers find AI comments useful while rejecting equivalence with human peer response.", "Read beside PAIRR: the key question includes who practices judgment and whether peer relationships remain central." ], "zh": [ "Warner 主张，同伴互评的重要学习发生在阅读别人、形成判断并反观自己的写作时。他承认有人觉得 AI 反馈有用，但反对把它视为与真人互评相同的学习活动。", "可与 PAIRR 对读：真正的分歧不只在反馈准不准，还在于谁得到练习判断的机会，以及同伴关系是否仍占有位置。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "warner-peer-response-20250328", "limits": { "en": "Normative argument and experience, not an outcome trial.", "zh": "规范性论证与教学经验；非效果比较试验。" }, "publishedAt": "2025-03-28", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "John Warner · Inside Higher Ed / Just Visiting", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://www.insidehighered.com/opinion/columns/just-visiting/2025/03/28/generative-ai-not-our-peer" } ], "stages": [ "higher" ], "summary": { "en": "Feedback quality is only part of peer response.", "zh": "反馈文本的质量不等于互评活动的全部价值。" }, "title": { "en": "AI Can’t Do Student Peer Review", "zh": "同伴互评的价值，首先在于给出反馈" }, "topics": [ "teaching", "assessment", "theory" ], "type": "opinion" }, { "body": { "en": [ "No More Marking combined teacher audio, AI-produced reports and student exercises. It abandoned automatic generation of highly specific questions because accuracy was insufficient, using human-written sets allocated by score instead.", "This exposes an important design trade-off: detailed personalization does not establish actionable feedback. Pupils can use paper, but the report supplies no independent learning comparison." ], "zh": [ "No More Marking 展示个人及班级反馈报告，组合教师音频、AI 整理和学生练习。系统最初想按每篇作文自动生成具体练习，但准确率不足，于是改用人工编写、按成绩分配的问题。", "这提供了可检查的失败记录：反馈篇幅和个性化程度不等于可行动性；精确度要求高的练习不能只依赖流畅生成。学生流程可以纸笔完成，但没有独立学习效果比较。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-feedback-design-20250321", "limits": { "en": "Commercial product-design account, not an efficacy trial or guarantee about the current version.", "zh": "产品设计说明，存在商业利益；不是效果试验，也不是当前版本保证。" }, "publishedAt": "2025-03-21", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/bringing-our-feedback-philosophy" } ], "stages": [ "school" ], "summary": { "en": "A design report explains why unreliable generated questions were replaced with human-written sets.", "zh": "设计复盘公开了 AI 出题失败后采用人工题库的取舍。" }, "title": { "en": "Bringing our feedback philosophy to life with AI", "zh": "反馈产品如何保留教师与学生的工作" }, "topics": [ "teaching", "assessment", "tools" ], "type": "practice" }, { "body": { "en": [ "The 2025 explanation separates learner capabilities from design decisions, informed by student concerns.", "It invites questioning whether AI is appropriate for the task. The current resource hub’s six-pillar revision should not be projected back onto this four-pillar account." ], "zh": [ "2025介绍区分学习者需要的批判能力与教师设计材料时要问的问题，并说明受到学生关注点启发。", "它鼓励按任务判断是否值得用 AI；后续资源页已扩为六支柱，不能把当前页面结构倒写为2025版本。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "ou-critical-ai-responsible-design-20250317", "limits": { "en": "Framework description, not evaluation; specific energy-use analogies were not validated.", "zh": "框架介绍，非效果验证；未接受文中的具体能耗类比为已核实数字。" }, "publishedAt": "2025-03-17", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Mirjam Hauck, Eleanor Moore & Mary Simper · OU Learning Design", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://www.open.ac.uk/blogs/learning-design/?p=2288" } ], "stages": [ "higher" ], "summary": { "en": "OU’s designers connect AI literacy with access, bias, environmental and labour questions.", "zh": "OU 设计者把公平、环境、偏差和劳动问题纳入课程选择，而不止于提示词技能。" }, "title": { "en": "Exploring two frameworks from The Open University: Critical AI Literacy Skills and Responsible by Design", "zh": "把批判性 AI 素养与负责任设计放进课程" }, "topics": [ "teaching", "policy", "theory" ], "type": "practice" }, { "body": { "en": [ "Students generated outlines, critiqued them in workshops, then completed the essay without further AI under course rules. Twenty joined the in-class poll.", "Their responses identify useful starting points but generic, similar and insufficiently detailed outputs." ], "zh": [ "学生先学习规则并生成提纲，再在课堂小组评价，之后按课程要求独立完善与写作。20名学生参加课堂意见调查。", "他们觉得提纲有助起步，却发现生成内容趋同、笼统并缺少专业深度。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "williams-ai-assessment-immunology-20250306", "limits": { "en": "Small descriptive case; no causal critical-thinking gain, and compliance outside class remains unverified.", "zh": "小样本描述性案例，不能证明批判思维增长；课外遵守限制仍是未验证前提。" }, "publishedAt": "2025-03-06", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Andrew Williams · Intersection / UCL Discovery", "scope": { "en": "Read repository PDF methods, teaching sequence, survey results and discussion; not the entire literature review.", "zh": "阅读机构PDF的方法、教学步骤、调查结果与讨论；非全部文献综述。" }, "status": "verified", "url": "https://discovery.ucl.ac.uk/id/eprint/10206017/" } ], "stages": [ "higher" ], "summary": { "en": "Williams describes staged AI outlining, peer critique and independent writing in an immunology course.", "zh": "Williams 展示一门硕士课程如何把 AI 生成提纲、同伴评价与独立写作分开。" }, "title": { "en": "Integrating Artificial Intelligence Into Higher Education Assessment", "zh": "从 AI 提纲到同伴批评：免疫学作业的重设计" }, "topics": [ "assessment", "teaching" ], "type": "practice" }, { "body": { "en": [ "Using repeated difficult-question trials, the authors show that prompt effects can reverse across questions and that evaluation thresholds alter conclusions about reliability.", "An editorial teaching idea is to have learners define reliability before comparing repeated outputs. This turns prompting advice into an inspectable experiment; that classroom application was not tested here." ], "zh": [ "团队用同一批难题反复测试 GPT-4o 系列，比较表达方式与判定标准。单题上的显著变化，可能在总体上相互抵消；平均分也可能掩盖不稳定。", "策展建议：把这篇当作 AI 素养的实验设计材料，让学习者先定义可靠性，再比较多次输出。论文测试的是模型表现，没有测试这项课堂活动。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "gail-prompt-contingency-20250304", "limits": { "en": "Model/benchmark-specific evidence. The suggested classroom use is an editorial inference.", "zh": "特定模型与 GPQA 题库的研究；课堂用途是策展推论，不是已验证的学习干预。" }, "publishedAt": "2025-03-04", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Lennart Meincke, Ethan Mollick, Lilach Mollick, Dan Shapiro · Wharton GAIL", "scope": { "en": "Read the public report’s methods and results; raw data and SSRN attachment not audited.", "zh": "阅读公开技术报告页的方法与结果；未审计原始数据或 SSRN 附件。" }, "status": "verified", "url": "https://gail.wharton.upenn.edu/research-and-insights/tech-report-prompt-engineering-is-complicated-and-contingent/" } ], "stages": [ "higher", "general" ], "summary": { "en": "Repeated tests illustrate why a successful answer does not establish a reliable prompting technique.", "zh": "Wharton 的重复测试提醒：一次好答案不足以证明某个提示词可靠。" }, "title": { "en": "Prompt Engineering is Complicated and Contingent", "zh": "AI 素养课可以教怎样检验提示词" }, "topics": [ "teaching", "tools" ], "type": "research" }, { "body": { "en": [ "He uses students’ beliefs about learning styles to illustrate why reported benefit can mislead.", "His proposal combines stakeholder voices with theory-informed investigation of learning processes and outcomes." ], "zh": [ "他以学生把个性化误解为匹配“学习风格”为例，说明自信的评价可能与学习证据脱节。", "文章要求建立理论驱动、情境明确的研究，既听学生意见，也检验更轻松的完成是否真的带来学习。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "lodge-real-learning-impact-20250209", "limits": { "en": "Authored commentary; referenced studies not all re-read, and rhetoric is not a universal causal claim.", "zh": "署名评论；引用的研究未全部重读，不把修辞当普遍因果结论。" }, "publishedAt": "2025-02-09", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Jason Lodge · Needed Now", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://needednowlt.substack.com/p/what-to-do-about-artificial-intelligence" } ], "stages": [ "higher" ], "summary": { "en": "Lodge calls for evidence about learning mechanisms beyond uptake and perceived usefulness.", "zh": "Lodge 主张把学生声音与学习过程、机制和更可靠的测量结合。" }, "title": { "en": "What to do about artificial intelligence now: Examining the real impact on learning", "zh": "研究 AI 学习影响，不能止于使用率与满意度" }, "topics": [ "learning", "theory" ], "type": "opinion" }, { "body": { "en": [ "Watkins examines his own distracted reading, allows a possible role for AI navigation, and worries about research outputs exceeding human verification capacity. He suggests protecting classroom time for sustained reading.", "For teaching, ask which original texts students must encounter and how much time and guidance the course actually provides." ], "zh": [ "Watkins 从自己读不完书和社交媒体分心的经历出发，区分浏览信息与深入阅读。他容许 AI 在资料导航中发挥作用，却担忧研究代理产出的来源规模超过人类核查注意力，并提出留出课堂阅读时间。", "值得带入教学的问题是：我们需要学生接触哪些原文，又给了他们多少真正阅读的时间与指导？" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "watkins-reading-social-media-20250126", "limits": { "en": "Personal reflection cannot establish population-level cognitive harm.", "zh": "个人反思；不能推出普遍的注意力或认知损害。" }, "publishedAt": "2025-01-26", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Marc Watkins · Rhetorica", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://marcwatkins.substack.com/p/reading-in-the-age-of-social-media" } ], "stages": [ "higher" ], "summary": { "en": "Reading habits are also an educator’s problem.", "zh": "阅读习惯的变化，不能只怪学生。" }, "title": { "en": "Reading in the Age of Social Media (and AI)", "zh": "教师自己的注意力困境，也是阅读教学的问题" }, "topics": [ "teaching", "learning", "theory" ], "type": "opinion" }, { "body": { "en": [ "Wetzel describes four-person board-game teams, assigned roles, AI-assisted rulebooks and peer playtesting followed by revision. A faculty workshop became a classroom experience; the aim was exploratory prototyping rather than production-ready output.", "Ask which decisions students originate and which they delegate. Human playtesting creates a useful reality check on generated designs." ], "zh": [ "Wetzel 介绍四人小组的桌游设计活动：成员分工，AI 协助更新规则，另一组试玩后反馈，再继续修改。它从教师工作坊发展成可供课程使用的体验；目标是探索设计，而非商业成品。", "阅读时可追问：哪些决定由学生提出，哪些由模型代做？真人试玩提供了一个让生成内容接受真实约束的环节。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "bruff-wetzel-design-accelerator-20250121", "limits": { "en": "Practice account; creativity transfer was not measured.", "zh": "实践说明；未测量创造力迁移。" }, "publishedAt": "2025-01-21", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Derek Bruff · Intentional Teaching", "scope": { "en": "Read prototyping, board-game and collaboration transcript sections.", "zh": "已读原型、桌游和协作转录段落。" }, "status": "verified", "url": "https://intentionalteaching.buzzsprout.com/2069949/episodes/16470938-ai-as-design-accelerator-with-ryan-wetzel" } ], "stages": [ "higher" ], "summary": { "en": "Scaffolding collaboration through roles and playtesting.", "zh": "用原型、角色与试玩组织人机协作。" }, "title": { "en": "AI as Design Accelerator with Ryan Wetzel", "zh": "让 AI 参与桌游设计，再由真人试玩检验" }, "topics": [ "teaching", "learning", "tools" ], "type": "practice" }, { "body": { "en": [ "Take-home work became effort/completion credit; major capability judgments moved into timed classroom exams with reattempts.", "The author explicitly identifies reduced scope for evaluation and creativity, unresolved online delivery, and the emotional cost of suspicion." ], "zh": [ "他把课外作业转为按完整性与努力计分，主要能力改由课堂限时考试验证，仍提供重考机会。", "他同时指出：这样挤压了评价与创造任务，也未解决异步在线课怎么办。文章的价值在于公开权衡，而非宣称找到了答案。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "talbert-ai-grading-triage-20250113", "limits": { "en": "One instructor’s account; suspicion is not verified misconduct. A historical 2025 response.", "zh": "单一教师自述；怀疑不等于确认作弊，方案是2025年当时的暂行安排。" }, "publishedAt": "2025-01-13", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Robert Talbert · Grading for Growth", "scope": { "en": "Read the original article body; not every linked resource.", "zh": "阅读原始文章正文；未审核全部外链。" }, "status": "verified", "url": "https://gradingforgrowth.com/p/how-ai-is-changing-my-grading-approach" } ], "stages": [ "higher" ], "summary": { "en": "Talbert documents an emergency redesign of discrete mathematics assessment and the uncertainty behind it.", "zh": "Talbert 记录离散数学课的应急改制，也承认无法测出真实 AI 作弊比例。" }, "title": { "en": "How AI is changing my grading approach -- for now", "zh": "AI 正在怎样改变我的评分方式：一份暂行方案" }, "topics": [ "assessment", "teaching" ], "type": "practice" }, { "body": { "en": [ "In 68 English secondary schools, 259 teachers joined a randomized trial. ChatGPT plus guidance reduced weekly Year 7–8 science planning from 81.5 to 56.2 minutes, a 25.3-minute difference.", "Use usually covered limited lesson components. Blinded resource review found no clear quality difference, but its sample was limited. Planning time is not student attainment or long-term learning." ], "zh": [ "英国 68 所中学的 259 位教师参与随机试验。使用 ChatGPT 加指导材料的一组，每周七、八年级科学备课平均 56.2 分钟，对照组 81.5 分钟，差 25.3 分钟、约 31%。", "教师多在一节课中的一两个环节使用。盲评未发现提交材料质量明显差异，但样本有限；研究重点是备课时间，不是课堂教学质量或学生长期成绩。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "eef-chatgpt-planning-rct-20241212", "limits": { "en": "RCT findings read from official results summaries; full evaluation report and appendices not audited.", "zh": "随机试验结论来自官方结果说明；本轮未完整读取评价报告和附录。" }, "publishedAt": "2024-12-12", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "EEF / NFER", "scope": { "en": "Read EEF’s dated result announcement, project summary and NFER research-director interpretation.", "zh": "已读 EEF 2024-12-12 结果正文、项目摘要和 NFER 研究负责人解读。" }, "status": "verified", "url": "https://educationendowmentfoundation.org.uk/news/teachers-using-chatgpt-alongside-a-guide-to-support-them-to-use-it-effectively-can-cut-lesson-planning-time-by-over-30-per-cent" }, { "checkedAt": "2026-10-07T09:44:18Z", "name": "NFER: Helen Poet", "scope": { "en": "Read the 18 March 2025 evaluator interpretation, particularly scope of use and quality-assessment limitations.", "zh": "已读 2025-03-18 研究者解读，尤其实际使用范围和质量评估局限。" }, "status": "verified", "url": "https://www.nfer.ac.uk/blogs/ai-help-or-hindrance/" } ], "stages": [ "school" ], "summary": { "en": "A 259-teacher trial supports planning-time savings without establishing higher student learning.", "zh": "259 位教师的试验支持特定备课环节减负，但没有证明学生学得更多。" }, "title": { "en": "ChatGPT in lesson preparation: the 2024 EEF/NFER trial", "zh": "科学备课随机试验：减负证据与学习证据分开" }, "topics": [ "teaching", "tools" ], "type": "research" }, { "body": { "en": [ "Warner distinguishes equal rules from meaningful learning: unrestricted AI access does not itself protect what credentials signify. He also questions surveillance costs and proxies such as page counts for academic rigor.", "Before drafting integrity rules, specify what students should gain and what evidence could credibly demonstrate it." ], "zh": [ "Warner 区分公平比较与真正学习：让所有人无限制用 AI，可能拉平规则，却不能自动保证课程证书仍有意义。他也批评监测挤占教学时间，以及把页数、字数当成严谨性的替代指标。", "这篇适合在制定诚信政策前阅读：先定义希望学生获得什么，再讨论允许范围和可接受证据。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "warner-integrity-llm-20241206", "limits": { "en": "Historical argument; detector predictions are not current facts.", "zh": "历史观点文；检测能力预测不作为当前事实。" }, "publishedAt": "2024-12-06", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "John Warner · Inside Higher Ed / Just Visiting", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://www.insidehighered.com/opinion/columns/just-visiting/2024/12/06/we-cant-let-chatgpt-make-lie-out-education" } ], "stages": [ "higher" ], "summary": { "en": "What should a course credential actually certify?", "zh": "从抓违规转向课程究竟证明什么。" }, "title": { "en": "Academic Integrity in an LLM World", "zh": "学术诚信不能只剩学生之间的公平竞争" }, "topics": [ "assessment", "policy", "theory" ], "type": "opinion" }, { "body": { "en": [ "Teachers recorded comments while comparing pupils’ writing; AI transcribed and combined them into individual and class reports. The Year 5 pilot involved 204 teachers in 31 schools and 5,321 comments.", "The design preserves human judgment while reducing production work. It leaves open whether pupils understand the feedback and improve later, independent writing." ], "zh": [ "教师比较学生作文时录入口头评论，AI 再把多位教师的评论转录、合并为学生评语及班级报告。作者称，31 所学校的 204 位教师在五年级项目中留下 5,321 条评论。", "价值在于清楚限定机器的工作：先收集教师对实际文本的判断，再减少整理时间。报告尚未证明学生读懂反馈、改变下一次写作或获得长期能力。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "nmm-human-voice-ai-comments-20241206", "limits": { "en": "Vendor-authored implementation report; no control group or transfer test.", "zh": "公司自己的实施报告；没有对照组或学习迁移测验。" }, "publishedAt": "2024-12-06", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Daisy Christodoulou / No More Marking", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://substack.nomoremarking.com/p/using-ai-to-automate-written-comments" } ], "stages": [ "school" ], "summary": { "en": "A product trial separates teachers’ evaluative judgment from AI-assisted feedback production.", "zh": "一项公开产品试验把评价判断留给教师，让 AI 转录并合并口头反馈。" }, "title": { "en": "Using AI to automate written comments", "zh": "先由教师判断，再让 AI 整理评语" }, "topics": [ "teaching", "assessment" ], "type": "practice" }, { "body": { "en": [ "Li sequences independent modeling before AI-supported extension and collects chats and reflections. She observes differing benefits by conceptual understanding, coding ability and AI literacy; some students find prompting no faster than working independently.", "When outputs improve, ask whether students gained conceptual understanding or merely bypassed implementation barriers. The distinction needs additional evidence." ], "zh": [ "Li 把机器学习作业分成不用 AI 和可用 AI 两段，收集对话及反思。她观察到概念知识、编程能力与 AI 素养共同影响获益；有学生认为写好提示不比自己完成更省时。", "这为课程设计提出具体问题：模型输出改善时，学生究竟补上了概念理解，还是只绕过实现障碍？需要额外证据区分。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "bruff-li-technical-courses-20241119", "limits": { "en": "Interview about 2023 practice; not a randomized trial.", "zh": "访谈回顾 2023 年课堂；非随机试验。" }, "publishedAt": "2024-11-19", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Derek Bruff · Intentional Teaching", "scope": { "en": "Read task design, observations and revision transcript sections.", "zh": "已读课程任务、观察与调整转录段落。" }, "status": "verified", "url": "https://intentionalteaching.buzzsprout.com/2069949/episodes/16129553-teaching-with-ai-in-technical-courses-with-jingjing-li" } ], "stages": [ "higher" ], "summary": { "en": "Independent modeling before AI-supported extension.", "zh": "先独立建模，再开放 AI，并检查过程。" }, "title": { "en": "Teaching with AI in Technical Courses with Jingjing Li", "zh": "概念基础怎样影响学生从 AI 编程中获益" }, "topics": [ "teaching", "learning", "assessment" ], "type": "practice" }, { "body": { "en": [ "Mills describes student-shared process reports with conferencing alternatives and no required full replay. She notes voice typing being mislabeled as pasted text and raises training-data settings.", "As a historical case, it prompts questions about diverse writing methods and whether alternatives to monitoring are genuinely usable." ], "zh": [ "Mills 记录 Grammarly Authorship 的课堂用法：学生分享过程报告，可另选面谈；她不要求完整回放，也指出语音输入被误标为粘贴。文章同时提醒考虑训练数据设置。", "作为历史案例，它适合追问过程证据是否覆盖不同写作方式，以及学生拒绝监测时是否有真实可行的替代路径。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "mills-authorship-beta-20241114", "limits": { "en": "2024 account, not a current software guide or endorsement.", "zh": "2024 年测试记录；不是当前软件指南或推荐。" }, "publishedAt": "2024-11-14", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Anna Mills · Substack", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://annamills.substack.com/p/trying-out-grammarly-authorship-beta" } ], "stages": [ "higher" ], "summary": { "en": "An early account of process evidence and privacy tradeoffs.", "zh": "一份早期实践记录，展示监测与隐私的取舍。" }, "title": { "en": "Trying out Grammarly Authorship Beta", "zh": "写作过程记录：提供选择，也保留误判警觉" }, "topics": [ "assessment", "tools", "teaching" ], "type": "practice" }, { "body": { "en": [ "Wexler connects writing-to-learn with retrieval and elaboration while recognizing that excessive writing demands can overload novices. She advocates explicit sentence and outline instruction embedded in curriculum content.", "This adds a useful premise to AI debates: preserving mental effort does not require unsupported struggle. The design problem is reducing unnecessary difficulty while retaining the thinking that teaches." ], "zh": [ "Wexler 把“写作促学”与检索练习、精加工研究连接起来，同时指出写作负担过高时，内容学习也可能受损。她主张明确教授句子与提纲，并让写作嵌入课程知识。", "这为 AI 写作争论补上前提：保留认知努力不等于让学生无支架地艰难写作。怎样减少无关负担，同时留下真正促学的思考，是更具体的设计问题。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "wexler-writing-effect-20241109", "limits": { "en": "Research interpretation, not a fresh study; references not fully re-audited. Author co-wrote The Writing Revolution.", "zh": "研究解读；未重新审读全部原研究。作者合著并支持 The Writing Revolution。" }, "publishedAt": "2024-11-09", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Natalie Wexler / Minding the Gap", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://nataliewexler.substack.com/p/the-writing-effect" } ], "stages": [ "school", "higher" ], "summary": { "en": "Writing’s retrieval and elaboration benefits must be balanced against novice cognitive load.", "zh": "写作的检索与精加工价值，必须与初学者的认知负荷一起考虑。" }, "title": { "en": "The Writing Effect", "zh": "写作为什么帮助学习，又为什么可能失效" }, "topics": [ "learning", "theory", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Watkins objects to hidden instructions designed to trap AI use. He raises transparency, assistive-reader exposure and students’ ability to challenge assessment methods; resistance to misuse should not excuse deception.", "A useful test is whether the assessment procedure can be openly explained, questioned and corrected." ], "zh": [ "Watkins 反对在作业中埋藏隐藏提示来抓学生。他强调透明性、辅助阅读技术可能读出隐藏内容，以及学生质疑评估方法的权利；反对 AI 滥用不能成为教学欺骗的理由。", "适合检视一个评估方案是否经得起公开解释：学生知道规则吗？教师的证据与判断过程能否被讨论和纠正？" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "watkins-deceptive-assessment-20241103", "limits": { "en": "Ethical argument; appeal predictions are not legal facts.", "zh": "伦理论证；不把作者关于申诉结果的预测当法律事实。" }, "publishedAt": "2024-11-03", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Marc Watkins · Rhetorica", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://marcwatkins.substack.com/p/deceptive-assessment-has-no-place" } ], "stages": [ "higher" ], "summary": { "en": "Assessment methods must meet their own integrity standard.", "zh": "把诚信要求同样施于教师的评估手段。" }, "title": { "en": "Deceptive Assessment Has No Place in Education", "zh": "教师用隐蔽陷阱抓 AI，也需要接受伦理检验" }, "topics": [ "assessment", "policy", "teaching" ], "type": "opinion" }, { "body": { "en": [ "Initially 425 students were randomised; platform analyses used 381 consenting students, with a smaller survey sample.", "Supportive feedback improved perceived usefulness and reduced reported anger, without a significant resource-quality difference. A tiny timing effect is not deeper learning." ], "zh": [ "425人最初参与，平台分析采用381名同意使用数据的学生；两组分别获得中性与情绪增强的 AI 反馈。", "问卷参与更少。创作任务的反馈后用时差异很小，不能将好感直接理解为更深学习。" ] }, "collectedAt": "2026-10-07T09:47:56Z", "id": "alsaiari-emotional-feedback-20241019", "limits": { "en": "Summarises the 2024 preprint; later journal version not fully compared, with no long-term retention or transfer measure.", "zh": "此条按2024预印本总结；后来期刊版本未全文比较，无长期保持或迁移测量。" }, "publishedAt": "2024-10-19", "sources": [ { "checkedAt": "2026-10-07T09:47:56Z", "name": "Omar Alsaiari, Nilufar Baghaei, Hatim Lahza, Jason Lodge, Marie Boden & Hassan Khosravi", "scope": { "en": "Read v1 methods, sample, results tables and conclusions; checked later journal listing, not a full version comparison.", "zh": "阅读v1方法、样本、结果表与结论；核对后续期刊题录，未逐表比对版本。" }, "status": "verified", "url": "https://arxiv.org/abs/2410.15077" } ], "stages": [ "higher" ], "summary": { "en": "A randomised feedback experiment separates improved perceptions from no detected improvement in work quality.", "zh": "随机分配试验发现更鼓励性的反馈评价较好、愤怒较少，但作品质量未见显著差异。" }, "title": { "en": "Emotionally Enriched Feedback via Generative AI", "zh": "让 AI 反馈更鼓励人：感受改变不等于作业质量提升" }, "topics": [ "learning", "assessment" ], "type": "research" }, { "body": { "en": [ "Mollick argues for teacher-led learning design and independent demonstrations of skill, rather than relying on confidence that AI use will be spotted.", "As an older reference, it helps trace the discussion’s starting problems. Its model capabilities, survey findings and detection claims must retain their 2024 context." ], "zh": [ "Mollick 主张让教师参与设计 AI 学习活动，重视学生能否独立思考，并讨论课堂展示与针对性的辅导。", "作为旧经典，它适合回看教育讨论的问题起点。文中的模型表现、调查与检测判断都有当时背景，应与较新的证据分开阅读。" ] }, "collectedAt": "2026-10-07T09:39:29Z", "id": "mollick-post-apocalyptic-education-20240830", "limits": { "en": "Outside the requested window; keep separate from the two-year and recent collections.", "zh": "明确早于2024-10-07窗口，不应混入近两年或近期栏目。" }, "publishedAt": "2024-08-30", "sources": [ { "checkedAt": "2026-10-07T09:39:29Z", "name": "Ethan Mollick · One Useful Thing", "scope": { "en": "Read the complete public essay; cited surveys and studies not individually re-evaluated.", "zh": "阅读全文；未逐一重评文中调查和研究。" }, "status": "verified", "url": "https://www.oneusefulthing.org/p/post-apocalyptic-education" } ], "stages": [ "school", "higher" ], "summary": { "en": "This 2024 essay distinguishes confidence in AI detection from the illusion that completed homework means learning.", "zh": "2024年的教育专文区分教师对检测的信心，与学生把完成作业误当作掌握知识的感觉。" }, "title": { "en": "Post-apocalyptic Education", "zh": "旧文重读：当作业答案随手可得" }, "topics": [ "assessment", "teaching", "learning" ], "type": "opinion" }, { "body": { "en": [ "UNESCO outlines 15 teacher competencies across five dimensions and three progression levels, supporting teacher development and assessment design while foregrounding ethics, human agency and pedagogy." ], "zh": [ "UNESCO 提出15项教师能力，涵盖以人为本、AI 伦理、基础与应用、AI 教学法及专业学习五个维度，分为获取、深化、创造三个层次，为教师培训与评价设计提供共同语言。" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "unesco-teacher-ai-framework-20240808", "limits": { "en": "A normative framework, not an implementation evaluation. Based on the official publication description, not a complete review of every language edition.", "zh": "规范性能力框架，不是实施效果评估；依据官方出版介绍，未声称完整审阅所有语言版本。" }, "publishedAt": "2024-08-08", "sources": [ { "checkedAt": "2026-10-07T07:32:44.824370+00:00", "name": "UNESCO", "scope": { "en": "Full official publication description read. Original publication: 8 August 2024; updated 16 January 2026, not treated as a new first publication.", "zh": "官方出版介绍全文已读。原发布2024年8月8日，更新2026年1月16日，不将更新当作首次发表。" }, "status": "verified", "url": "https://www.unesco.org/en/articles/ai-competency-framework-teachers" } ], "stages": [ "school", "higher" ], "summary": { "en": "UNESCO outlines 15 teacher competencies across five dimensions and three progression levels, supporting teacher development and assessment design while foregrounding ethics, human agency and pedagogy.", "zh": "UNESCO 提出15项教师能力，涵盖以人为本、AI 伦理、基础与应用、AI 教学法及专业学习五个维度，分为获取、深化、创造三个层次，为教师培训与评价设计提供共同语言。" }, "title": { "en": "AI competency framework for teachers", "zh": "UNESCO 教师 AI 能力框架：从工具操作走向伦理与教学判断" }, "topics": [ "teaching", "policy", "theory" ], "type": "policy" }, { "body": { "en": [ "Anti-cheating measures can undermine assessment when their effects exclude students or distort evidence.", "This frames later reform debates and is retained separately as an out-of-window foundation." ], "zh": [ "一项反作弊措施即使降低作弊，也可能因为排斥部分学生而损害评价本身。", "此文解释了后续评价改革的核心问题意识，作为窗口外的历史阅读线索保存。" ] }, "collectedAt": "2026-10-07T09:51:22Z", "id": "dawson-validity-over-cheating-20240808", "limits": { "en": "Historical August 2024 conceptual paper outside the main window; full text not read in this pass.", "zh": "2024年8月的历史概念文，超出主窗口；本轮未读全文。" }, "publishedAt": "2024-08-08", "sources": [ { "checkedAt": "2026-10-07T09:51:22Z", "name": "Phillip Dawson, Margaret Bearman, Mollie Dollinger & David Boud · UTS repository", "scope": { "en": "Read repository abstract/metadata and checked publisher/Crossref first-online date; PDF fetch unsuccessful.", "zh": "只读机构摘要与元数据，并核对出版社和Crossref的首次在线日期；PDF读取未成功。" }, "status": "limited", "url": "https://opus.lib.uts.edu.au/handle/10453/187950" } ], "stages": [ "higher" ], "summary": { "en": "The authors place cheating within the broader question of defensible judgments about capability.", "zh": "作者把作弊问题纳入更大的效度问题：毕业能力判断是否可信。" }, "title": { "en": "Historical foundation: Validity matters more than cheating", "zh": "历史基础：评价效度比抓作弊更重要" }, "topics": [ "assessment", "theory" ], "type": "opinion" }, { "body": { "en": [ "Wexler inspects commercially offered companions to her book, finding errors and missing narrative detail. She asks what readers lose when summaries replace the original.", "This May 2024 piece provides historical context for her later learning critique. It is firsthand inspection of specific materials, not a learning experiment or proof that summaries are never useful." ], "zh": [ "Wexler 检视市场上未经自己参与编写的《The Knowledge Gap》配套读物，指出错误、人物混淆以及情境和细节被削去。她由此追问，摘要能否替代原书带来的理解。", "这篇 2024 年 5 月的旧文帮助理解她后续的学习论批评。它是作者对具体材料的第一手观察；不意味着所有摘要都无用，也不是学生学习效果试验。" ] }, "collectedAt": "2026-10-07T09:44:18Z", "id": "wexler-ai-summaries-foundational-20240505", "limits": { "en": "Predates the two-year window and belongs only in historical context; concerns specific products.", "zh": "早于两年窗口，单独历史区保存；本文描述的是特定摘要产品。" }, "publishedAt": "2024-05-05", "sources": [ { "checkedAt": "2026-10-07T09:44:18Z", "name": "Natalie Wexler / Minding the Gap", "scope": { "en": "Read byline, date and article body; not every external reference was independently checked.", "zh": "已读署名、日期和文章正文；未独立核验全部外链。" }, "status": "verified", "url": "https://nataliewexler.substack.com/p/ai-generated-summaries-are-a-problem" } ], "stages": [ "general", "school" ], "summary": { "en": "An author’s inspection of generated book companions reveals losses beyond factual accuracy.", "zh": "作者检视自己书籍的生成式配套摘要，展示准确性之外的阅读损失。" }, "title": { "en": "Historical reference: AI-Generated Summaries Are a Problem—for Authors and Society as a Whole", "zh": "历史参考：AI 摘要替代阅读时，遗漏了什么？" }, "topics": [ "learning", "theory" ], "type": "opinion" }, { "body": { "en": [ "Watkins draws on a classroom pilot to recognize access possibilities while questioning speed-focused substitutes for close reading. He proposes annotation and discussion and records differing student responses.", "Read with his 2026 reading project to trace the move from identifying a problem to designing an alternative environment." ], "zh": [ "Watkins 从阅读助手课堂试用出发，承认语言与可及性潜力，同时担忧速度导向替代细读。他提出用社会化批注与讨论保留阅读中的停顿和判断，并记录学生对省时的不同反应。", "可将其与 2026 年阅读项目连读，观察同一位教师如何从提出风险，转向尝试设计替代环境。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "classic-watkins-ai-reading-20240503", "limits": { "en": "Not controlled research; historical product details are not current facts.", "zh": "非对照研究；历史产品描述不当作当前事实。" }, "publishedAt": "2024-05-03", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "Marc Watkins · Rhetorica", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://marcwatkins.substack.com/p/no-one-is-talking-about-ais-impact" } ], "stages": [ "higher" ], "summary": { "en": "Earlier classic: reading assistance versus reading replacement.", "zh": "窗口外经典：便利与深读之间的早期警讯。" }, "title": { "en": "No One is Talking About AI’s Impact on Reading", "zh": "经典回看：当 AI 替学生阅读，究竟省掉了什么？" }, "topics": [ "teaching", "learning", "theory" ], "type": "opinion" }, { "body": { "en": [ "Warner recalls heavily annotated portfolios left uncollected and questions feedback that mainly justifies grades. He favors genuine reader response and opposes automated writing assessment.", "This earlier argument provides a clear counterpoint to AI-feedback practice. Both approaches must confront what students understand and do with comments." ], "zh": [ "Warner 回顾自己花时间批注、学生却不领取作品集的经历。他把问题归于反馈主要在解释分数，转而强调真实读者的回应，并坚持反对让 AI 承担写作评价。", "它与支持 AI 反馈的实践构成清楚的思想分歧。共同的检验起点是学生如何理解并采取行动，而不只是评论由谁生产。" ] }, "collectedAt": "2026-10-07T09:51:04Z", "id": "classic-warner-meaningful-feedback-20240205", "limits": { "en": "2024 opinion; experience does not prove effectiveness.", "zh": "2024 年观点；经验不构成效果证明。" }, "publishedAt": "2024-02-05", "sources": [ { "checkedAt": "2026-10-07T09:51:04Z", "name": "John Warner · Inside Higher Ed / Just Visiting", "scope": { "en": "Public article body read.", "zh": "已读公开正文。" }, "status": "verified", "url": "https://www.insidehighered.com/opinion/columns/just-visiting/2024/02/05/ai-cant-read-why-pay-attention-its-feedback-writing" } ], "stages": [ "higher" ], "summary": { "en": "Earlier classic: ask whether feedback is actually used.", "zh": "窗口外经典：先问反馈是否真正被使用。" }, "title": { "en": "On AI and ‘Meaningful’ Feedback", "zh": "经典回看：反馈到底是为了打分，还是为了写作者？" }, "topics": [ "assessment", "teaching", "theory" ], "type": "opinion" }, { "body": { "en": [ "Matuschak distinguishes forgetting from never understanding, and discusses questioning, memory and fading scaffolds. The interview offers a lens for AI tools: do they help learners notice confusion and act, or mainly ease consumption?" ], "zh": [ "Matuschak 区分“读过后忘记”与“从未真正理解”，讨论提问、回忆练习和撤去学习支架。他提醒工具设计不能只让内容更易消费。一个值得用于 AI 助手的问题是：它是否帮助学习者察觉困惑、行动并获得反馈？" ] }, "collectedAt": "2026-10-07T07:35:55.125056+00:00", "id": "dwarkesh-matuschak-learning-tools-20230712", "limits": { "en": "Research-informed views and observations, not a general effectiveness result. This 2023 discussion is not current AI capability reporting; the AI connection is editorial.", "zh": "受访者的研究知情观点与观察，不是一般效果结论。2023年访谈不能被当作当前 AI 能力报道；AI 联系是编辑的阅读问题。" }, "publishedAt": "2023-07-12", "sources": [ { "checkedAt": "2026-10-07T07:32:44.834480+00:00", "name": "Dwarkesh Podcast · Andy Matuschak", "scope": { "en": "Full official written transcript read, focusing on learning. Audio and YouTube video not played. Publisher chapters: 00:06:52 self-teaching, 00:16:37 memory, 00:44:27 learning media.", "zh": "已读官方完整文字转录，摘要聚焦学习部分。未播放音频或YouTube视频；章节时间00:06:52自学、00:16:37记忆、00:44:27学习媒介。" }, "status": "verified", "url": "https://www.dwarkesh.com/p/andy-matuschak" } ], "stages": [ "higher", "school", "general" ], "summary": { "en": "Matuschak distinguishes forgetting from never understanding, and discusses questioning, memory and fading scaffolds. The interview offers a lens for AI tools: do they help learners notice confusion and act, or mainly ease consumption?", "zh": "Matuschak 区分“读过后忘记”与“从未真正理解”，讨论提问、回忆练习和撤去学习支架。他提醒工具设计不能只让内容更易消费。一个值得用于 AI 助手的问题是：它是否帮助学习者察觉困惑、行动并获得反馈？" }, "title": { "en": "Andy Matuschak — The reason most learning tools fail", "zh": "Andy Matuschak：学习工具应让人发现不理解之处" }, "topics": [ "theory", "learning", "tools" ], "type": "opinion" } ] }