Paper List
-
Macroscopic Dominance from Microscopic Extremes: Symmetry Breaking in Spatial Competition
This paper addresses the fundamental question of how microscopic stochastic advantages in spatial exploration translate into macroscopic resource domi...
-
Linear Readout of Neural Manifolds with Continuous Variables
This paper addresses the core challenge of quantifying how the geometric structure of high-dimensional neural population activity (neural manifolds) d...
-
Theory of Cell Body Lensing and Phototaxis Sign Reversal in “Eyeless” Mutants of Chlamydomonas
This paper solves the core puzzle of how eyeless mutants of Chlamydomonas exhibit reversed phototaxis by quantitatively modeling the competition betwe...
-
Cross-Species Transfer Learning for Electrophysiology-to-Transcriptomics Mapping in Cortical GABAergic Interneurons
This paper addresses the challenge of predicting transcriptomic identity from electrophysiological recordings in human cortical interneurons, where li...
-
Uncovering statistical structure in large-scale neural activity with Restricted Boltzmann Machines
This paper addresses the core challenge of modeling large-scale neural population activity (1500-2000 neurons) with interpretable higher-order interac...
-
Realizing Common Random Numbers: Event-Keyed Hashing for Causally Valid Stochastic Models
This paper addresses the critical problem that standard stateful PRNG implementations in agent-based models violate causal validity by making random d...
-
A Standardized Framework for Evaluating Gene Expression Generative Models
This paper addresses the critical lack of standardized evaluation protocols for single-cell gene expression generative models, where inconsistent metr...
-
Single Molecule Localization Microscopy Challenge: A Biologically Inspired Benchmark for Long-Sequence Modeling
This paper addresses the core challenge of evaluating state-space models on biologically realistic, sparse, and stochastic temporal processes, which a...
Autonomous Agents Coordinating Distributed Discovery Through Emergent Artifact Exchange
Laboratory for Atomistic and Molecular Mechanics (LAMM), Massachusetts Institute of Technology | Department of Biological Engineering, MIT | Department of Mechanical Engineering, MIT | Department of Civil and Environmental Engineering, MIT | Department of Materials Science and Engineering, MIT | Center for Computational Science and Engineering, Schwarzman College of Computing, MIT
30秒速读
IN SHORT: This paper addresses the fundamental limitation of current AI-assisted scientific research by enabling truly autonomous, decentralized investigation where multiple AI agents coordinate without central planning through emergent artifact exchange.
核心创新
- Methodology Introduces ArtifactReactor for plannerless coordination using pressure-based scoring (novelty, centrality, depth, age) to prioritize needs fulfillment and schema-overlap matching for multi-parent synthesis across independent analyses.
- Methodology Implements a comprehensive artifact layer with immutable records, SHA-256 content hashes, and DAG-based provenance tracking that enables full computational lineage from raw outputs to published findings.
- Methodology Develops a persistent ecosystem with autonomous mutation layer that actively prunes redundant workflows and resolves conflicts, plus persistent memory allowing agents to build upon complex epistemic states across multiple cycles.
主要结论
- The framework successfully demonstrated heterogeneous tool chaining across 300+ interoperable scientific skills spanning biology, materials science, chemistry, and genomics domains.
- Four autonomous investigations showed emergent convergence among independently operating agents, with cross-domain applications including peptide design for SSTR2 receptor and resonance bridging across biology, materials, and music.
- The system enables traceable reasoning from raw computation to published findings through comprehensive provenance tracking in artifact DAGs, creating auditable scientific records with full computational lineage.
摘要: We present ScienceClaw + Infinite, a framework for autonomous scientific investigation in which independent agents conduct research without central coordination, and any contributor can deploy new agents into a shared ecosystem. The system is built around three components: an extensible registry of over 300 interoperable scientific skills, an artifact layer that preserves full computational lineage as a directed acyclic graph (DAG), and a structured platform for agent-based scientific discourse with provenance-aware governance. Agents select and chain tools based on their scientific profiles, produce immutable artifacts with typed metadata and parent lineage, and broadcast unsatisfied information needs to a shared global index. The ArtifactReactor enables plannerless coordination: peer agents discover and fulfill open needs through pressure-based scoring, while schema-overlap matching triggers multi-parent synthesis across independent analyses. An autonomous mutation layer actively prunes the expanding artifact DAG to resolve conflicting or redundant workflows, while persistent memory allows agents to continuously build upon complex epistemic states across multiple cycles. Infinite converts these outputs into auditable scientific records through structured posts, provenance views, and machine-readable discourse relations, with community feedback steering subsequent investigation cycles. Across four autonomous investigations, peptide design for the somatostatin receptor SSTR2, lightweight impact-resistant ceramic screening, cross-domain resonance bridging biology, materials, and music, and formal analogy construction between urban morphology and grain-boundary evolution, the framework demonstrates heterogeneous tool chaining, emergent convergence among independently operating agents, and traceable reasoning from raw computation to published finding.