Cross-disciplinary collection
AI4Science field
Research releases and public discussion in this scientific field, ordered by time.
Large language model agents can improve across tasks by retaining reusable skills distilled from prior interactions. Recent work jointly optimizes task execution and skill extraction, enabling the policy and skillbank to co-evolve. However, as the actor continues learning, rewarding skill proposals through their reuse in subsequent training steps may conflate skill benefits with actor improvement, while directly test
This work examines the transfer of a co-evolved communication mechanism between two robotic agents from a discrete two-dimensional (2D) simulator to a three-dimensional simulator with real physics (3D). The study focuses on whether a communication mechanism co-evolved in a 2D environment retains its functional role after transfer to a 3D physics-based simulator. To support this analysis, the effects of the episode ti
Dual-process agents pair a fast policy with a slow deliberative model. In real-time settings the slow model usually runs continuously; in turn-based agents and robot planners it is invoked on events such as uncertainty or a detected failure. We study a fast learned actor that takes every decision and hands control to a reasoning vision-language model only when a gate opens, while the game keeps running. We use closed
Latent world models have shown a remarkable ability to predict future states and to plan in the real world. In practice, however, we lack a principled way to estimate how their capabilities scale with model size, data, and compute, an open problem that slows progress in the field. In this work we present RoboJEPA, a world model based on the Joint Embedding Predictive Architecture (JEPA) and trained on a large-scale d
Modern information systems, including many agentic workflows, use dense retrieval to explore large amounts of unstructured data. However, dense retrieval relies on surface-level semantic similarity, which is insufficient for increasingly complex search applications. Here, we investigate agentic retrieval that combines the reasoning capabilities of Large Language Models (LLMs) with the efficient corpus exploration of
A large language model (LLM) agent solves long-horizon tasks through many reasoning-action turns, with one verification signal at termination. Deployed agents face streams of related tasks, making their trajectories a natural resource for improvement. In-context adaptation agents store reflections, memories, or skills as text, so reuse depends on retrieving the right experience and on a frozen policy executing it. We
LLM-based agents are increasingly capable of generating complex 3D structures, with the potential to reshape how objects are designed and realized in the physical world. Yet, producing elegant geometry is fundamentally different from producing objects that can be built and perform their intended functions. Existing evaluations largely focus on geometric quality while overlooking physical realizability. We introduce L
Terminal agents are increasingly deployed beyond software engineering in science and other specialized domains. Constructing training environments requires executable reference behavior and a domain-specific verifier that distinguishes semantic correctness from superficially plausible artifacts. Authoring these components for each task requires repeated engineering and limits reuse. We introduce software-in-the-loop
Vision-Language-Action (VLA) foundation models have scaled rapidly to enhance manipulation performance and generalizability, but this scaling incurs high computational costs that render real-world deployment increasingly challenging. Existing approaches typically mitigate this issue by designing smaller architectures or reducing the iterative denoising steps in flow-based policies. In this work, we propose FastOPD, a
World action models integrate future visual dynamics with robot action prediction, but their scalability remains limited by the need for action-annotated robot trajectories. Observation-only videos contain rich evidence about interaction dynamics, but existing approaches typically use them either to pretrain visual representations that must later be adapted for control, or to infer latent actions that are subsequentl
Computer-use agents need to reliably ground action targets in complex desktop scenes, where multiple applications, overlapping windows, and visually similar controls compete for attention. Existing training data rarely pair such scenes with dense annotations or vary them in a controlled way. We introduce DeskForge, a controllable desktop environment that composes and explores real applications to generate large-scale
What makes great scientists great? Even as AI systems start to make progress on open problems, scientists remain far ahead of them at sensing which prior idea, buried in an ever-growing archive of research, a new problem needs. To study this skill, we draw on researchers who know firsthand which earlier work advanced their completed projects, with papers serving as pointers to the ideas within. Using our automated pi
As LLM agents undertake increasingly complex, long-horizon tasks, verifying their outputs becomes increasingly challenging. We study how verification capability can be strengthened with a fixed base model, without access to reference answers or grading rubrics at test time. Repeated sampling yields multiple rollouts that can contain complementary correct claims, but we need a reliable verification mechanism to determ
LLM agents generate intermediate reasoning and actions token by token, making extended interactions slow and computationally expensive. Jev-style models offer fast probabilistic predictions over finite fields, but require those fields to be specified in advance. This requirement limits autonomous task solving, where the available actions must be derived from natural language instructions and adapted through interacti
This work evaluates the direct transfer of a co-evolved communication protocol from a 2D simulation to a 3D physical environment, without retraining the network weights. Two e-puck-type robots, controlled by a GRU network with residual connection, were evaluated in a food-seeking task with social signaling. The sensory and motor translation layer required three corrections for stable physical operation, including the
Manipulation behaviors vary widely across objects and scenes, but they share a small set of reusable skills, and planning with these skills helps embodied agents generalize to new tasks. Yet an agent can only plan with skills it knows. Recovering skills from observed experience, the inverse of planning, builds this knowledge over time and yields skill data for training future agents. Vision-Language Models (VLMs) des
Coding agents are beginning to move beyond purely digital tasks to tackle physical-world challenges, particularly in robotics. Existing robotics benchmarks, however, primarily focus on the performance of individual artifacts, such as policies or controllers, offering limited coverage of coding agents' broader engineering capabilities. Real-world robotics extends beyond control: agents must build, integrate, diagnose,
Reinforcement learning has helped AI agents solve increasingly difficult tasks, but high rewards do not always reflect the work users intended. In recent incidents and controlled evaluations across the AI industry, agents trained to maximize reward have accessed unauthorized information, attempted to evade monitoring systems, and even breached sandbox protections to attack external systems. As agents become more capa
Before invoking external tools, an agentic LLM must select among a K-way action space: executing a call, seeking clarification, answering directly, or declining. While internal activation steering can alter these pre-execution decisions, conventional aggregate metrics obscure where altered states land and what collateral damage they inflict. We present SAKIKO, an auditing framework that formalizes representation repa
Multi-step agents are trained on flat action streams: SFT and RLVR weight every token uniformly and ignore the sub-procedures that recur across tasks, the hierarchy that lets humans plan top-down from reusable routines. This structure sits unused, and flat training uses each scarce trajectory less fully than its content allows. Recent agents do use that structure, but only as LLM-written skills in context, never in t
First up on the podcast, Staff Reporter Tom Howarth joins host Sarah Crespi to talk about the language artificial intelligence agents evolve inside their swarms . Next on the show, producer Elah Feder talks with Revathe Thillaikumar , a behavioral ecologist and postdoctoral fellow at the Max Planck Institute of Animal Behavior. The two discuss the value of social learning as baby orangutans learn to find their own fo
Tool-calling LLM agents are increasingly deployed in enterprise applications. However, effective evaluation and optimization require high-quality, diverse task datasets that are often difficult to obtain due to privacy and other constraints. Existing synthetic task generation methods often produce generic tasks that ignore an agent's underlying state or database and fail to reflect real-world usage diversity. We prop
LLM agents are increasingly deployed in collaborative settings, yet long-term interaction may give rise to undesirable coordination. We study the emergence of collusion in a long-horizon multi-agent environment: two agents repeatedly complete individual tasks, share task logs, verify each other's work, and receive rewards. We introduce realistic constraints that make compliance with the verification protocol incompat
Skills can improve the performance of Large Language Model (LLM) agents by providing task-specific procedural guidance, while skill optimization further improves their effectiveness through iterative refinement. However, existing skill optimization methods typically represent skills as unstructured natural-language instructions, creating two key challenges: 1) Unstructured skills often lack explicit workflow-level gu
AI models need to do more than produce correct answers. How they respond matters too: whether they’re helpful, fair, safe, respectful, and responsive to the people using them. For model builders, the challenge is knowing whether those “prosocial” behaviors hold up in practice—and whether evaluations capture how a model behaves when people interact with it in unexpected ways. Northeastern University MS student Soham P
ScienceIDE: Turning World's Scientific Codebase into Agent Learnable Environments 摘要 Scientific code repositories encode decades of human knowledge in executable models, methods, and tools. Yet fragmented toolchains, implicit domain conventions, and specialized correctness criteria make this knowledge difficult to convert into reliable learning experience—a challenge we call the scientific experience bottleneck. We i
Spatial intelligence requires more than describing object locations. Under incomplete observation, models must identify and acquire missing evidence, interpret it in a common spatial frame, and act on it. We introduce VA-Bench to evaluate the complete observe-reason-act-revise loop. General-purpose MLLMs learn procedural context from RGB-only demonstrations, actively select camera viewpoints, issue metric Cartesian c
Enabling robots to adapt to unfamiliar environments as readily as humans remains a moonshot goal of embodied AI. No finite collection of demonstrations can cover every task and situation a robot will encounter, making the ability to learn from context at deployment essential for generalization. Such in-context learning (ICL), however, remains largely beyond the reach of existing robotic policies. The broad agentic ca
ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents 摘要 ScienceBuddy is an interactive scientific research workspace that brings continually improving agents into a researcher's everyday work. It supports researchers in carrying out scientific tasks while turning their requests, feedback and execution evidence into tasks and evaluation rubrics for continual learning. At its core is
Scientific research is a continuous process that emphasizes inheritance. Methods developed by predecessors are often expanded upon by new researchers to explore more novel and in-depth scientific questions. However, the change of lab staff, such as student graduation, leads to a lack of personnel capable of replicating methods. Methods that have been developed with significant effort and resources cannot be continued
Large Language Model (LLM) agents offer a promising path toward autonomously managing long-term physical tasks without human intervention. However, physical tasks require agents to continuously observe the environment, make consequential actions, and remain effective as the environment changes. Existing approaches either require substantial data and retraining, or primarily focus on agents operating in the virtual wo
LLM assistants are widely used for daily social advice, yet evaluating their social reasoning in such consultation settings remains challenging since (i) it requires setups where the assistant learns about social situations from subjective user narratives, and (ii) social properties, such as others' intentions, typically lack verifiable ground truth. To address these challenges, we introduce Fuse, a multi-agent simul
PhAI Labs announced its DFM research framework, published the DFM technical report and research-map repository, and said ScienceBuddy, ScienceIDE and JEPA-Anything were scheduled as separate releases on September 16–18.
Discovery Foundation Models: Toward Open-Ended Discovery Intelligence 摘要 Foundation models have progressed from learning and reasoning over existing knowledge, to increasingly learning through action, tool use, and outcome feedback. We argue that the next frontier is a further transition: from solving and acting within problems specified by humans to participating in the process by which new problems, representations
Machine learning, at its core, is about generalization, not memorization. You hand your learning algorithm a pile of training examples and use them to fit a model. But the goal is not to perform well on the training examples — that's easy, you could just memorize the answers. The goal is to perform well on new examples that you have never before seen. If a model does well on the data it was trained on but poorly on f
Digital agents must often adapt to new environments whose interfaces, tools, and failure modes are not fully captured by pretrained models. We introduce RSIAgent, a training-free multi-agent framework for recursive self-improvement through autonomous memory construction. RSIAgent coordinates curriculum, actor, and verifier agents to continually explore the environment, validate outcomes, and retain environment-specif
When people engage in spontaneous thought under low external constraint, thought content changes continuously and can shift across topics. However, the cognitive computational principles governing these transitions remain unclear, as does whether similar dynamics emerge in artificial language generation. Here, we formalized spontaneous narrative dynamics using a foraging-inspired framework, constructing an agent base
Visual goal and dynamics prediction can provide language-conditioned robot policies with both a target outcome and a representation of action-dependent scene changes. We bring these predictions into action generation and selection through a shared trajectory model. Dynin-Robotics implements this formulation on Dynin-Omni, an omnimodal masked-diffusion backbone, representing language, visual observations, goals, and a
We demonstrate that a general-purpose agent can directly drive a physical robot throughout task execution without any task-specific or environment-specific training. We introduce Agent as Policy (AGP), which places task planning and execution under the agent's control. Given a task and a robot interface, the agent interprets visual evidence, writes executable programs, issues motion commands, and revises its actions
In this technical report, we propose Pelican-Sim 1.0, a general world model simulator for embodied intelligence that predicts future observations from visual context and robot actions to support downstream learning and decision making. The model incorporates four key design features: (1) Unified action representation: a 28-dimensional action value space covering most mainstream embodiments, keeping one model valid ac
A research idea may be novel, coherent, and scientifically plausible, yet its proposed method may remain insufficiently specified for faithful implementation. We study the codification readiness of implementation-facing research-method specifications, defined by whether they provide sufficient methodological information for a competent implementer or coding agent to construct the intended method without unsupported a
The ability of AI systems to improve their behavior during deployment is becoming increasingly important. As inference moves beyond the static execution of a fixed trained model, a growing body of work studies how models can refine their behavior on the fly by exploiting test-time information and additional computation. These developments have largely evolved along two directions: methods that modify the model's stat
Black-box optimization problems remain challenging because of large, weakly structured, and high-dimensional search spaces. Existing methods often suffer from poor sample efficiency because they rely on direct candidate generation or trial-and-error refinement. A natural way to improve search efficiency is to use world modeling, which can help identify promising optimization directions before costly evaluation. Large
Scientific analysis workflows encode deep domain knowledge through sequences of tightly coupled operations where correctness depends on tool selection, execution order, and parameterization. A CFD engineer investigating flow separation must extract wall shear stress, identify zero-crossings in skin friction, and confirm with boundary-layer profiles: a chain that requires both domain expertise and proficiency with vis
Scientific progress depends not only on finding solutions, but on learning the rules that explain why they work and using that understanding to design better experiments. We introduce science sandboxes, a framework for studying this capability in AI agents through repeated cycles of experimentation, feedback, and hypothesis revision. Science sandboxes invite an agent to query the natural world in different ways, rang
Large Language Model (LLM) agents have shown promise for automating data-science workflows, yet their end-to-end performance depends critically on the agent harness that represents tasks, manages execution state, constrains output artifacts, and provides evaluation feedback. Existing data-science agents often leave this harness implicit, making results difficult to reproduce, compare, and attribute across heterogeneo
Large language model (LLM) agents are increasingly embedded in scientific workflows for literature analysis, drafting, and review. Existing systems advance autonomous discovery and manuscript generation, but do not resolve the governance problem that arises when ideas, methods, results, and claims propagate through AI-assisted workflows without mandatory human approval or artifact-level traceability. This paper propo
Heuristic search underlies planning in autonomous systems ranging from warehouse logistics to robotic navigation, yet generic heuristics fail to exploit the structural constraints that govern constrained spatial domains, causing search performance to degrade catastrophically on harder instances. We study this problem through the two-column Flying Block Puzzle, a rigorously NP-complete spatial planning microworld whos
Graphical user interfaces have emerged as an important environment for evaluating autonomous AI agents on multimodal interactive tasks. Existing benchmarks such as AndroidWorld and MobileWorld provide strong foundations for mobile agent evaluation, but their application coverage and task design do not yet fully capture the diversity and complexity of realistic mobile use. We present GMA, a benchmark for evaluating ge
Natural language driven autonomous co-scientist workflows involve a fundamental trade-off between flexibility and reasoning at the expense of determinism, reproducibility, and observability. Such agents increasingly must communicate across institutional boundaries, where federation topology can shape latency and cost. We systematically evaluated these tradeoffs using a controlled ablation on a production agentic plat
A trained reinforcement learning policy does not determine the complete behavior that users encounter: deployment code still schedules, admits, suppresses, or replaces its proposed actions. We contribute \emph{runtime action interference} (RAI), an AI control mechanism that preserves policy parameters while regulating action pacing and filtering configured action patterns after inference. RAI releases a proposed acti
Literature reviews are essential to scientific progress, but rigorously evaluating automatically generated reviews remains difficult because many aspects of research utility depend on expert judgment rather than reference-overlap metrics. We introduce LitReview Arena, a battle-style evaluation platform with a structured protocol tailored to literature review quality: domain experts with AI paper-writing experience co
Advances in large language models (LLMs) have fueled a wave of research into agency: the ability to reason, plan, and act. This effort has produced agentic frameworks that orchestrate perception, memory, and decision-making around powerful LLM backbones. With the advent of large multimodal models (LMMs), these systems can process and integrate diverse modalities, including images, audio, and video, thereby improving
The high photon flux at synchrotron beamlines allows for the measurement of fast dynamical processes. However, beamline radiation-safety protocols prohibit human intervention during X-ray experiments, limiting the ability to perform versatile real-time sample manipulations during continuous data acquisition. Here we present a robotic platform at an X-ray scattering beamline to enable real-time sample handling and pro
Training terminal agents requires scalable executable supervision, yet synthesizing high-quality terminal tasks is challenging. Each task couples an instruction, an initialized environment, a reference solution, and an executable verifier; inconsistent assumptions can make tasks unsolvable or mis-evaluated. Multi-stage synthesis may discard original goals, dependencies, state transitions, and procedural constraints.
ASI-Bench is the first benchmark to jointly evaluate AI systems' capabilities in innovative exploration and autonomous scientific execution across general research domains, and the first to progressively withdraw human methodological guidance within the same project to test how far AI can proceed independently. Built by over 40 experts with 31,000+ human hours, it includes 60 project-level tasks across 11 scientific
Reinforcement learning (RL) has shown promise in single-turn LLM fine-tuning, but long-horizon agentic reasoning introduces increasingly branching interactions and sparse rewards, exposing limitations: heavyweight backpropagation-based training stacks and harder credit assignment. This paper argues that evolution strategies (ES) are a better choice, offering model scalability (full-parameter optimization with inferen
We present RapidLiDAR, a LiDAR scene completion method that treats initialization as a learned, data-driven component. An adaptive initialization module predicts spatially varying displacement for each partial input point, avoiding manual noise tuning. A multi-scale reconstruction module refines point positions using voxel and BEV feature maps. On SemanticKITTI and KITTI-360, our method achieves state-of-the-art comp
Prime Intellect ran 153 internet-isolated autonomous runs across 18 frontier models on the nanoGPT optimizer speedrun, releasing the traces, monitor reports, per-run ledger and harness.
LLM-based agents act on behalf of users to access cloud services, call tools, or invoke agents, but permissions remain static and each request is evaluated independently, allowing deviations from delegated tasks, prohibited combinations of actions, or unrestricted sub-delegation. The Agentic Principal Chain (APC) tracks delegated authority and evaluates requests against accumulated session state using six checks, car
AI co-scientists are beginning to transform research by generating hypotheses, retrieving related work, designing experiments, executing code, and drafting papers, but current systems ignore the individual researcher, optimizing generic metrics like novelty and reviewer scores. This paper introduces personalized auto-research, which conditions every stage on a representation of the researcher, arguing that personaliz
AutoResearchEval introduces 100 tasks spanning 7 scientific domains and the full research lifecycle, including ideation, retrieval, execution, analysis, writing, and review. Evaluating 8 harness-model combinations yields 800 trajectories with process-level annotation, leading to the AutoResearch Failure Taxonomy (ARFT), a framework of 45 empirically-grounded failure patterns. A human-calibrated agent-as-a-judge pipel
LLM coding agents issue Bash commands through interfaces that may serialize, wrap, and reparse model output. Matched execution scores alone cannot distinguish command-generation errors from failures introduced after generation. QuoteBench measures this boundary with exact final-state validation on 56 one-shot tasks from 14 incident-derived families, crossing the generation contract with the execution transport around
Existing AI scientist systems typically reason over text, code, or precomputed summaries, missing crucial spatial, temporal, and cross-channel evidence. To address this, researchers introduce OmniScientist, an end-to-end omni-modal AI scientist that directly conducts multidisciplinary research from heterogeneous raw evidence. The system features a perception layer and three autonomous agents for ideation, experiment,
AINews reports on a new paper demonstrating how encrypted reasoning traces from frontier models can be decoded and replayed to other models, sessions, or users, leading to dramatic improvements in open models. Since the release of o1, reasoning models have obscured their traces with cryptographic signatures to prevent distillation, but this study shows that these encrypted blocks can be replayed to weaker models and
Biomni integrates tool retrieval, code execution and experimental protocols across biomedical tasks in a single agent environment.
Abstract Self-driving lab systems (aka, autonomous experimentation) accelerate research - letting scientists learn faster, spend less resources, and fail smarter in well defined, narrow studies. The next-generation materials lab combines self-driving systems to tackle broader challenges - orchestrating complex research campaigns while optimizing lab resources. We propose that agent-based and agentic artificial intell
A Stanford-led research team that includes Fei-Fei Li published a benchmark on May 18, 2026, that documents a specific, measurable failure in every frontier multimodal AI model tested: when forced to actively move through a 3D environment rather than interpret a pre-composed image, state-of-the-art systems consistently choose the wrong actions, skip viewpoints that would correct their mistakes, and commit to wrong an
FutureHouse's head of science explains its agent stack, a drug-repurposing case, and why scientific publishing and evaluation must be redesigned for machine readers.