AutoScientists
Free while signed in. Answers cite the passages they came from.
AutoScientists, from Harvard, is a decentralized team of AI agents for long-running computational science that drops the central planner entirely. Rather than following one research trajectory coordinated from the top, agents self-organize around promising hypotheses, critique each other's proposals before spending experimental compute, and record both successes and failures so the system avoids redundant exploration as evidence accumulates over hours or days.
No central planner: Agents interpret shared experimental state, form teams around promising directions, and reorganize when progress stalls. Coordination emerges from a common state rather than a top-level controller, which sustains parallel search instead of a single thread.
Evaluate before you spend: Proposals are critiqued and scored before any experimental compute is allocated. This gating reduces wasted trials and keeps the system from repeating dead ends that an individual agent would otherwise revisit.
Strong results on real science tasks: On BioML-Bench, 24 biomedical ML tasks spanning imaging, protein engineering, single-cell omics, and drug discovery, AutoScientists reaches 74.4% mean leaderboard percentile, an improvement of +8.33% over the strongest prior AI agent.
Why it matters: Most multi-agent research systems still funnel decisions through a planner that becomes a bottleneck. Decentralized self-organization with explicit failure-sharing is a different blueprint for long-horizon scientific search, and it holds up on hard biomedical benchmarks.
Get next week’s papers.
The same picks and the same summaries, in your inbox. Free, and 176 issues deep.
Subscribe on Substack