Research index
Automatically collected and categorized papers from arXiv, upstream awesome lists, and other research sources.
ReasFlow is a multi-agent system for autonomous reasoning-centric scientific discovery in applied mathematics, integrating verification, knowledge retrieval, and paper generation.
AI CFD Scientist is an open-source AI agent that autonomously conducts computational fluid dynamics discovery with physics-aware verification, achieving a 7.89% improvement in wall Cf RMSE.
ReasFlow is a multi-agent system for reasoning-centric scientific discovery in applied mathematics, automating proofs, synthesis, and paper generation.
Presents a framework for fully automated scientific discovery using LLMs to generate ideas, write code, run experiments, write papers, and simulate review.
Introduces DISCOVERYWORLD, a virtual environment with 120 tasks for benchmarking automated scientific discovery agents.
Introduces an AI system that autonomously conducts research and writes papers, achieving the first AI-generated workshop paper accepted in peer review.
AutoResearchClaw is a multi-agent autonomous research pipeline with human-AI collaboration that outperforms AI Scientist v2 by 54.7% on ARC-Bench.
Introduces BAISBench, a benchmark evaluating AI scientists on real single-cell transcriptomic data for cell annotation and biological discovery.
Claw AI Lab is an interactive multi-agent autonomous research platform with customizable roles, real-time monitoring, and a code harness for experiments, evaluated on AI case studies.
A comprehensive survey of agentic AI systems for scientific discovery, covering progress, challenges, and future directions across multiple scientific fields.
Case study of four autonomous ML research paper attempts using LLM agents, documenting failure modes and design principles.
SAGE uses multi-hypothesis failure attribution to self-correct autonomous research agents, boosting success rates on a 12-topic benchmark.
Proposes a negative knowledge memory layer for AI research agents that records failures, improving performance on scientific tasks while using fewer tokens.
Introduces AblationBench, a benchmark for evaluating LM agents on planning ablation experiments in AI research, finding current LMs underperform humans.
Jr. AI Scientist autonomously generates novel research papers from baseline papers by analyzing limitations, experimenting, and writing, achieving higher review scores.
EvoMaster is a domain-agnostic, self-evolving agent framework for scalable agentic science, achieving state-of-the-art results on four benchmarks.
CausalForge automates theoretical research in causal inference using a Lean proof assistant and a self-improving agentic pipeline.
Proposes an autonomous recommender-systems research lab (AutoRecLab) integrating end-to-end automation and outlines an agenda for building, evaluating, and governing such labs.
OR-Agent is a multi-agent framework combining evolutionary search and structured research trees for automated algorithm discovery.
ARISE uses specialized LLM agents and rubric-guided iterative refinement to automatically generate high-quality academic survey papers.
Kosmos is an AI scientist that autonomously performs cycles of data analysis, literature search, and hypothesis generation over 12 hours, producing traceable scientific reports with high accuracy.
A domain-specific multi-agent system (EDM-ARS) automates end-to-end educational data mining research, producing LaTeX manuscripts with validated analyses.
CARTOGRAPH is a verification layer for AI scientists that uses experiment steering, ambiguity closure, and refusal to decide when to stop, validated across multiple testbeds and real-world audits.
DeepScientist autonomously discovers scientific findings via Bayesian optimization and hierarchical validation, surpassing human SOTA on three AI tasks.
SAGE uses multi-hypothesis failure attribution to self-correct autonomous research agents, improving success rates and artifact quality on a 12-topic benchmark.
MLIPilot uses LLM agents to autonomously optimize machine-learned interatomic potentials via hypothesis proposal, code editing, and HPC job management.
MIND is an LLM-driven multi-agent framework for automated hypothesis validation in materials research using in-silico experiments.
FARS is a fully automated AI research system that produced 166 papers across 67 topics, evaluated by 282 structured reviews.
First benchmark evaluating academic integrity in AI scientist systems, finding 34.2% failure rate across 7 LLMs due to completion bias.
An LLM pipeline autonomously conducts literature review, reproduces experiments, performs novel computations, and writes a manuscript in condensed-matter physics with fault tolerance.
AI-Supervisor is a multi-agent framework with a persistent Research World Model for autonomous end-to-end AI research supervision.
A survey of AI-powered research automation, defining AutoResearch and Vibe Research, analyzing workflow conditions from literature to reporting.
FirstResearch introduces a structured Research Question Certificate for LLM scientific discovery agents, enabling auditable question formation and outperforming baselines in preliminary evaluations.
SCION is an agentic scientific operating system that coordinates tasks, tools, and agents via a Research Execution Plan for long-horizon discovery.
Argues for developing multi-agent AI scientists within reformed institutions, illustrated by the Denario prototype, to accelerate discovery.
Proposes an iterative workflow for automated literature survey generation using recurrent outline generation and paper cards, outperforming baselines.
LABBench2 is a benchmark of nearly 1,900 biology tasks measuring real-world AI capabilities, showing increased difficulty over its predecessor.
SR-Scientist uses LLMs as autonomous agents to discover scientific equations by writing code, analyzing data, and optimizing equations, outperforming baselines across four disciplines.
Evaluation of Sakana's AI Scientist reveals critical flaws: poor literature reviews, high experiment failure rate, minimal code changes, and hallucinated results.
Presents The AI Scientist, an end-to-end system that autonomously conducts AI research from idea to publication, passing peer review at a workshop.
SMCEvolve uses Sequential Monte Carlo to guide LLM-driven program evolution for principled scientific discovery with convergence guarantees.
AlphaResearch is an autonomous agent that discovers new algorithms via iterative proposal, programming, and optimization using dual rewards from execution and simulated peer review.
Robin is a multi-agent system that fully automates scientific discovery, identifying a novel treatment for dry AMD.
Identifies four failure modes in AI scientist systems and demonstrates them via controlled experiments on two open-source systems.
Xcientist externalizes research synthesis and validation into inspectable processes, preserving traceable trajectories and addressing claim drift in automated AI research.
AutoSOTA is an end-to-end multi-agent system that automates replication and optimization of SOTA AI models, discovering 105 new SOTA models across top conferences.
AutoSurvey2 automates literature survey generation using retrieval-augmented synthesis, iterative refinement, and multi-LLM evaluation, outperforming baselines.
BioResearcher is an end-to-end LLM-based multi-agent system automating dry lab biomedical research, achieving 63% success on eight novel objectives.