AutoScientists: Self-Organizing Agent Teams for Long-Running Scientific Experimentation

2026-05-27

AI Summary

Current AI-driven scientific research systems struggle to support parallel exploration, dynamic adaptation, and systematic logging of failed trajectories in long-term experiments. This work proposes the first decentralized multi-agent framework for scientific discovery, enabling sustainable and collaborative automation through self-organized coordination, hypothesis-driven experimental planning, critical proposal evaluation, and cross-task knowledge sharing. By overcoming the limitations of single-trajectory exploration and centralized scheduling, the framework demonstrates consistent performance gains: it outperforms the strongest baseline by an average of 8.33% on BioML-Bench, accelerates GPT training optimization by 1.9× while identifying seven effective improvements, and achieves a 6.5%–12.5% increase in Spearman correlation on ProteinGym.

Abstract

Scientific research proceeds through iterative cycles of hypothesis generation, experiment design, execution, and revision. AI agents can automate parts of this process, but existing approaches typically follow a single research trajectory or coordinate through a central planner with fixed objectives. As a result, they struggle to sustain parallel exploration, adapt as experimental evidence changes, or preserve knowledge of failed directions over long-running experiments. We introduce AutoScientists, a decentralized team of AI agents for long-running computational scientific experimentation. Agents interpret a shared experimental state, self-organize into teams around promising hypotheses, critique proposals before using experimental compute, and share successes and failures to reduce redundant exploration. Under matched experimental budgets, AutoScientists improves over prior AI agents across biomedical machine learning, language-model training optimization, and protein fitness prediction. On BioML-Bench, spanning biomedical imaging, protein engineering, single-cell omics, and drug discovery, AutoScientists achieves a mean leaderboard percentile of 74.4% across 24 tasks, improving over the strongest AI agent by +8.33%. On GPT training optimization, AutoScientists reaches a target validation bits-per-byte 1.9x faster than Autoresearch and continues discovering improvements from a starting champion where the single-agent approach finds none (7 vs. 0 accepted improvements). On ProteinGym fitness prediction, AutoScientists discovers a method for ACE2-Spike binding that improves over the current state-of-the-art model by +12.5% in Spearman correlation. Applied without modification across all 217 ProteinGym assays, the same method improves over the prior state of the art by +6.5% (Spearman correlation).

Problem

• long-running scientific experimentation
• parallel exploration
• adaptive hypothesis revision
• knowledge preservation
• decentralized coordination

Innovation

• decentralized AI agents
• self-organizing teams
• long-running scientific experimentation
• hypothesis-driven exploration
• knowledge sharing