AI reviews for all 22,977 AAAI-26 papers were preferred by authors and PC members over human reviews on accuracy and suggestions and outperformed baselines at spotting weaknesses.
super hub Canonical reference
The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
Canonical reference. 75% of citing Pith papers cite this work as background.
223
Pith papers citing it
107
external citations · Pith
Background
75% of classified citations
abstract
One of the grand challenges of artificial general intelligence is developing agents capable of conducting scientific research and discovering new knowledge. While frontier models have already been used as aides to human scientists, e.g. for brainstorming ideas, writing code, or prediction tasks, they still conduct only a small part of the scientific process. This paper presents the first comprehensive framework for fully automatic scientific discovery, enabling frontier large language models to perform research independently and communicate their findings. We introduce The AI Scientist, which generates novel research ideas, writes code, executes experiments, visualizes results, describes its findings by writing a full scientific paper, and then runs a simulated review process for evaluation. In principle, this process can be repeated to iteratively develop ideas in an open-ended fashion, acting like the human scientific community. We demonstrate its versatility by applying it to three distinct subfields of machine learning: diffusion modeling, transformer-based language modeling, and learning dynamics. Each idea is implemented and developed into a full paper at a cost of less than $15 per paper. To evaluate the generated papers, we design and validate an automated reviewer, which we show achieves near-human performance in evaluating paper scores. The AI Scientist can produce papers that exceed the acceptance threshold at a top machine learning conference as judged by our automated reviewer. This approach signifies the beginning of a new era in scientific discovery in machine learning: bringing the transformative benefits of AI agents to the entire research process of AI itself, and taking us closer to a world where endless affordable creativity and innovation can be unleashed on the world's most challenging problems. Our code is open-sourced at https://github.com/SakanaAI/AI-Scientist
hub tools
citation-role summary
background 31
baseline 2
dataset 1
method 1
other 1
citation-polarity summary
claims ledger
- abstract One of the grand challenges of artificial general intelligence is developing agents capable of conducting scientific research and discovering new knowledge. While frontier models have already been used as aides to human scientists, e.g. for brainstorming ideas, writing code, or prediction tasks, they still conduct only a small part of the scientific process. This paper presents the first comprehensive framework for fully automatic scientific discovery, enabling frontier large language models to perform research independently and communicate their findings. We introduce The AI Scientist, which
authors
co-cited works
representative citing papers
AutoLab benchmark shows frontier models mostly fail at sustained iterative optimization due to premature termination, with persistence as the key success factor.
Agent-BRACE improves LLM agent performance on long-horizon partially observable tasks by 5.3-14.5% through a decoupled belief state of verbalized atomic claims with certainty labels that keeps context length constant.
A new native-runtime benchmark reveals that current frontier AI agents succeed on at most 62 percent of realistic long-horizon CLI tasks.
AutoResearchBench is a new benchmark showing top AI agents achieve under 10% success on complex scientific literature discovery tasks that demand deep comprehension and open-ended search.
Introduces ARA as a four-layer machine-executable research package and reports benchmark gains in agent QA accuracy and reproduction success.
FermiLink is a unified AI agent framework that automates multidomain scientific simulations via separated package knowledge bases and a four-layer progressive disclosure mechanism, reproducing 56% of target figures in benchmarks and generating research-grade results on unpublished problems.
A Lean-verified multi-agent system produces a catalogue of 14,116 quantum codes with transversal diagonal gates for small parameters, extracts infinite families, and resolves specific distance-3 cases with constructions and no-go proofs.
ZendoWorld shows that high labeling accuracy does not equal rule recovery, perception and induction are separate bottlenecks, and VLM agents propose near-uninformative experiments on active visual concept induction.
LLM-generated research ideas cluster more around bridge-like opportunities and synthesis methods than the broader distribution seen in human papers.
Multi-agent LLM system Agora under Sealed Joint Search conditions produces +1.87 holdout Sharpe on CSI 1000 over a 91-day sealed period, exceeding the best baseline at +1.334 under favorable seed.
The paper proposes AICID as a new identifier system to make the provenance of AI-generated scholarly work transparent and machine-readable.
Glite ARF introduces a verifier-driven three-role framework for parallel LLM coding agents, demonstrated by first- and second-place finishes in the BEA 2026 vocabulary-difficulty shared task across three languages with 29.9-35.9% RMSE reduction at ~$450 API cost.
A token-based authorization system with simulation and human gates enables safe LLM-written ARTIQ control code execution on trapped-ion platforms while blocking unauthorized hardware access.
Frontier coding agents surpass published Nature-family SOTA on only 17.8% of 90 sealed scientific tasks, mostly by recasting problems as supervised ML rather than inventing methods.
PixJail automates construction of paper-specific attack modules and unified evaluation pipelines for text-to-image jailbreaks, reproducing eleven methods with 2.1% average and 0% median error.
An LLM-driven agent with built-in seed-noise audits develops control policies for two aerospace problems that outperform undirected search and pass verification checks.
ENPIRE supplies four modules (Environment, Policy Improvement, Rollout, Evolution) that turn real-world robot training into an autonomous optimization loop driven by coding agents.
ICL-derived intrinsic rewards are biased in general MDPs but asymptotically match true learning progress in non-temporal settings, with supporting experiments.
PhySciBench benchmark shows current AI models achieve at most 33.5% accuracy on physical science tasks; DelveAgent framework improves accuracy by up to 7.5 points and cuts costs to one-third.
SafeClawBench supplies 600 staged adversarial tasks and three separate endpoints that show semantic acceptance, audit evidence, and sandbox-observed harm are distinct failure modes in tool-using LLM agents.
MetaSyn is a stage-level benchmark of 442 meta-analyses showing LLM agents retrieve up to 90.9% of eligible studies but include at most 52.7% in their final reports.
T2SP converts time series into structured programs for trends, periods, and events, enabling off-the-shelf LLMs to perform better on editing, captioning, and QA tasks than raw string inputs.
LLM agents match or exceed human methodological diversity and produce aligned effect estimates, yet flip final verdicts from 10% to 90% support under a confirmatory prompt while leaving coefficients unchanged.
citing papers explorer
-
AI-Assisted Peer Review at Scale: The AAAI-26 AI Review Pilot
AI reviews for all 22,977 AAAI-26 papers were preferred by authors and PC members over human reviews on accuracy and suggestions and outperformed baselines at spotting weaknesses.
-
AutoLab: Can Frontier Models Solve Long-Horizon Auto Research and Engineering Tasks?
AutoLab benchmark shows frontier models mostly fail at sustained iterative optimization due to premature termination, with persistence as the key success factor.
-
Agent-BRACE: Decoupling Beliefs from Actions in Long-Horizon Tasks via Verbalized State Uncertainty
Agent-BRACE improves LLM agent performance on long-horizon partially observable tasks by 5.3-14.5% through a decoupled belief state of verbalized atomic claims with certainty labels that keeps context length constant.
-
WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation
A new native-runtime benchmark reveals that current frontier AI agents succeed on at most 62 percent of realistic long-horizon CLI tasks.
-
AutoResearchBench: Benchmarking AI Agents on Complex Scientific Literature Discovery
AutoResearchBench is a new benchmark showing top AI agents achieve under 10% success on complex scientific literature discovery tasks that demand deep comprehension and open-ended search.
-
The Last Human-Written Paper: Agent-Native Research Artifacts
Introduces ARA as a four-layer machine-executable research package and reports benchmark gains in agent QA accuracy and reproduction success.
-
FermiLink: A Unified Agent Framework for Multidomain Autonomous Scientific Simulations
FermiLink is a unified AI agent framework that automates multidomain scientific simulations via separated package knowledge bases and a four-layer progressive disclosure mechanism, reproducing 56% of target figures in benchmarks and generating research-grade results on unpublished problems.
-
Co-Designing Quantum Codes with Transversal Diagonal Gates via Multi-Agent Systems
A Lean-verified multi-agent system produces a catalogue of 14,116 quantum codes with transversal diagonal gates for small parameters, extracts infinite families, and resolves specific distance-3 cases with constructions and no-go proofs.
-
Playing ZendoWorld: Challenging AI Agents on Active Visual Concept Induction
ZendoWorld shows that high labeling accuracy does not equal rule recovery, perception and induction are separate bottlenecks, and VLM agents propose near-uninformative experiments on active visual concept induction.
-
Measuring the Gap Between Human and LLM Research Ideas
LLM-generated research ideas cluster more around bridge-like opportunities and synthesis methods than the broader distribution seen in human papers.
-
AI Trading's Alpha Singularity: Emergent Market Reasoning through Agent-to-Agent Self-Evolution
Multi-agent LLM system Agora under Sealed Joint Search conditions produces +1.87 holdout Sharpe on CSI 1000 over a 91-day sealed period, exceeding the best baseline at +1.334 under favorable seed.
-
AICID: Unique Identifiers for AI Scientists
The paper proposes AICID as a new identifier system to make the provenance of AI-generated scholarly work transparent and machine-readable.
-
Glite ARF: Verifier-Driven Research with Parallel LLM Coding Agents
Glite ARF introduces a verifier-driven three-role framework for parallel LLM coding agents, demonstrated by first- and second-place finishes in the BEA 2026 vocabulary-difficulty shared task across three languages with 29.9-35.9% RMSE reduction at ~$450 API cost.
-
A hardware-safety-gated system for LLM-written native ARTIQ control code on a trapped-ion platform
A token-based authorization system with simulation and human gates enables safe LLM-written ARTIQ control code execution on trapped-ion platforms while blocking unauthorized hardware access.
-
NatureBench: Can Coding Agents Match the Published SOTA of Nature-Family Papers?
Frontier coding agents surpass published Nature-family SOTA on only 17.8% of 90 sealed scientific tasks, mostly by recasting problems as supervised ML rather than inventing methods.
-
PixJail: Self-Evolving Paper-to-Pipeline Reproduction for Text-to-Image Jailbreak Evaluation
PixJail automates construction of paper-specific attack modules and unified evaluation pipelines for text-to-image jailbreaks, reproducing eleven methods with 2.1% average and 0% median error.
-
Agentic AutoResearch forSpace Autonomy: An Auditable, LLM-Driven Research Agent for Aerospace Control Problems
An LLM-driven agent with built-in seed-noise audits develops control policies for two aerospace problems that outperform undirected search and pass verification checks.
-
ENPIRE: Agentic Robot Policy Self-Improvement in the Real World
ENPIRE supplies four modules (Environment, Policy Improvement, Rollout, Evolution) that turn real-world robot training into an autonomous optimization loop driven by coding agents.
-
Can In-Context Learning Support Intrinsic Curiosity?
ICL-derived intrinsic rewards are biased in general MDPs but asymptotically match true learning progress in non-temporal settings, with supporting experiments.
-
Deep Research in Physical Sciences: A Multi-Agent Framework and Comprehensive Benchmark
PhySciBench benchmark shows current AI models achieve at most 33.5% accuracy on physical science tasks; DelveAgent framework improves accuracy by up to 7.5 points and cuts costs to one-third.
-
SafeClawBench: Separating Semantic, Audit-Evidence, and Sandbox Harm in Tool-Using LLM Agents
SafeClawBench supplies 600 staged adversarial tasks and three separate endpoints that show semantic acceptance, audit evidence, and sandbox-observed harm are distinct failure modes in tool-using LLM agents.
-
MetaSyn: A Benchmark for LLM Agents on Meta-Analysis Articles from Nature Portfolio
MetaSyn is a stage-level benchmark of 442 meta-analyses showing LLM agents retrieve up to 90.9% of eligible studies but include at most 52.7% in their final reports.
-
Representing Time Series as Structured Programs for LLM Reasoning
T2SP converts time series into structured programs for trends, periods, and events, enabling off-the-shelf LLMs to perform better on editing, captioning, and QA tasks than raw string inputs.
-
AI Coding Agents in Social Science: Methodologically Diverse, Empirically Consistent, Interpretively Vulnerable
LLM agents match or exceed human methodological diversity and produce aligned effect estimates, yet flip final verdicts from 10% to 90% support under a confirmatory prompt while leaving coefficients unchanged.
-
Data Journalist Agent: Transforming Data into Verifiable Multimodal Stories
Data2Story is a multi-agent framework that generates evidence-grounded multimodal articles from data, evaluated on 18 articles against human pieces for verifiability, angle coverage, and quality across human, rubric, and automated judges.
-
Harnessing the Collective Intelligence of AI Agents in the Wild for New Discoveries
EinsteinArena is a platform for AI agents to collectively discover new mathematical results through open interaction, achieving 12 new state-of-the-art outcomes including raising the 11-dimensional kissing number lower bound from 593 to 604.
-
InquiTree: Evaluating AI Agents in the Scientific Inquiry Loop with Paper-Derived Research Trees
InquiTree shows LLM agents suffer from degrading critical capabilities during extended scientific interactions and perform worse on papers published after their training cutoffs.
-
TianJi-Environ: An Autonomous AI Scientist for Atmospheric Environmental Research
TianJi-Environ is a WRF-Chem-based multi-agent AI framework for autonomous validation of atmospheric chemistry mechanisms through executable experiments and evidence assessment.
-
AutoMedBench: Towards Medical AutoResearch with Agentic AI Models
AutoMedBench evaluates AI agents on long-horizon medical workflows across five stages and finds validation and submission as dominant failure points based on thousands of runs.
-
DiscoverPhysics: Benchmarking LLMs for Out-of-the-Box Scientific Thinking
DiscoverPhysics is a new benchmark with 22 on-demand N-body simulated worlds where LLM agents design experiments to infer non-standard physics, evaluated via held-out trajectory MSE and LLM-judged explanation quality.
-
LLM-driven design of physics-constrained constitutive models: two agents are better than one
A Creator-Inspector multi-agent LLM pipeline for constitutive artificial neural networks increases the rate of models satisfying all nine physical constraints to 100% or 56% depending on the LLM backbone.
-
Whose Good, Whose Place? The Moral Geography of Agentic AI for Social Good
Survey of 112 agentic AI for social good papers reveals moral-geographic asymmetry with 73% lacking geographic context (lowest for SDG 16) and only 25% reporting deployments.
-
IdleSpec: Exploiting Idle Time via Speculative Planning for LLM Agents
IdleSpec improves LLM agent accuracy by generating and aggregating speculative plans during idle time between tool calls and observations using complementary drafting strategies.
-
FML-bench: A Controlled Study of AI Research Agent Strategies from the Perspective of Search Dynamics
FML-Bench shows a simple greedy hill-climber nearly matches tree search on dense-opportunity tasks while an adaptive agent that broadens search on stagnation outperforms six baselines across 18 tasks.
-
1GC-7RC: One Graphic Card -- Seven Research Challenges! How Good Are AI Agents at Doing Your Job?
Introduces the 1GC-7RC benchmark to evaluate AI coding agents on seven diverse ML tasks under single-GPU time and access constraints.
-
The Scientific Contribution Graph: Automated Literature-based Technological Roadmapping at Scale
Builds a 2M-contribution graph from 230k papers with 12.5M prerequisite links and reports 0.48 MAP on temporal backtesting for predicting enabling technologies.
-
Graphs of Research: Citation Evolution Graphs as Supervision for Research Idea Generation
GoR extracts citation DAGs using position, frequency, predecessor links and time, then fine-tunes Qwen2.5-7B on 498 seed papers to generate ideas, claiming SOTA over gpt-4o baselines via LLM judges.
-
Sheaf-Theoretic Transport and Obstruction for Detecting Scientific Theory Shift in AI Agents
A finite sheaf-theoretic framework ranks obstruction measures to identify when an AI agent's theory must deform within its language or extend to a new one, validated on a controlled transition benchmark.
-
Harnessing Agentic Evolution
AEvo introduces a meta-agent that edits the evolution procedure or agent context based on accumulated state, outperforming baselines by 26% relative improvement on agentic benchmarks and achieving SOTA on open-ended tasks.
-
ASIA: an Autonomous System Identification Agent
ASIA uses an LLM-based coding agent to autonomously perform system identification, tested empirically on two benchmarks while noting limitations in transparency and reproducibility.
-
PaperFit: Vision-in-the-Loop Typesetting Optimization for Scientific Documents
PaperFit uses rendered page images in a closed loop to diagnose and repair typesetting defects in LaTeX documents, outperforming baselines on a new benchmark of 200 papers.
-
Hypothesis-Driven Deep Research with Large Language Models: A Structured Methodology for Automated Knowledge Discovery
HDRI is a six-principle eight-stage framework for hypothesis-organized LLM research featuring gap-driven iteration, traceable fact reasoning, and subject locking, realized in INFOMINER with reported gains in fact density and completeness.
-
MLS-Bench: A Holistic and Rigorous Assessment of AI Systems on Building Better AI
Current AI agents fail to reliably invent transferable ML methods on a controlled 140-task benchmark, and more test-time compute or context does not remove the scientific-judgment bottleneck.
-
Self-Driving Datasets: From 20 Million Papers to Nuanced Biomedical Knowledge at Scale
Starling, a multi-agent LLM system, extracts ~6.3 million nuanced structured records from PubMed across six tasks with reported error rates of 0.6-7.7%, lower than several curated databases.
-
AI co-mathematician: Accelerating mathematicians with agentic AI
An interactive AI workbench for mathematicians achieves 48% on FrontierMath Tier 4 and helped solve open problems in early tests.
-
AI CFD Scientist: Toward Open-Ended Computational Fluid Dynamics Discovery with Physics-Aware AI Agents
AI CFD Scientist autonomously discovers a Spalart-Allmaras runtime correction reducing lower-wall Cf RMSE by 7.89% on the periodic hill at Reh=5600 while using a vision-language gate to detect 14 of 16 silent failures missed by solver checks.
-
Evolving Idea Graphs with Learnable Edits-and-Commits for Multi-Agent Scientific Ideation
EIG represents research ideas as evolving graphs with nodes for claims and edges for relations, using a learned controller for edits and commits to produce higher-quality scientific proposals than text-only multi-agent baselines.
-
Agentic-imodels: Evolving agentic interpretability tools via autoresearch
Agentic-imodels evolves scikit-learn regressors via an autoresearch loop to jointly boost predictive performance and LLM-simulatability, improving downstream agentic data science tasks by up to 73% on the BLADE benchmark.
-
Theory Under Construction: Orchestrating Language Models for Research Software Where the Specification Evolves
Comet-H orchestrates LLMs via deficit-scoring prompt selection and half-life task tracking to co-evolve research software components, demonstrated by a static analysis tool reaching F1=0.768 versus a 0.364 baseline.
-
End-to-end autonomous scientific discovery on a real optical platform
An LLM agent autonomously identifies and experimentally validates a previously unreported optical bilinear interaction on a physical platform.