Agent swarms can already do a large part of routine theoretical research, and the scarce resource…
agent swarms can already do a large part of routine theoretical research, and the scarce resource flipped from producing results to reviewing them: an arxiv perspective argues trust, credit and incentives are the open problems now that human input is often just a prompt. deepmind’s co-scientist, published in nature, runs closed-loop experiments in materials and biology.
the question isn’t whether ai can research, it’s who gets paid when compute does the work.
Context
The arXiv paper 'AI Agent Swarms as Researchers' (Sep 28, 2026) says swarms of off-the-shelf coding agents, given a short statement of scope, can carry out long multi-step research with little supervision. A separate arXiv paper, 'A Society of Researchers' (Oct 7, 2026), proposes institutions for populations of autonomous research agents.
Google DeepMind's co-scientist work was published in Nature on May 19, 2026 (DeepMind LinkedIn post); The Decoder (Aug 28) says it now plans experiments, runs lab equipment and writes papers.
The agent swarm paper and the Nature publication match the sources. The Nature date is May 19, 2026, which predates the arXiv perspective. The 'closed-loop experiments in materials and biology' wording was not seen in the sources read, so unsupported here, not refuted. for the Nature paper itself; The Decoder describes lab equipment and papers. Which arXiv perspective the post means is not confirmed, and the 'trust, credit and incentives' framing was not seen in the sources read, so unsupported here, not refuted. 'Who gets paid when compute does the work' is the author's.
Related work
- arXiv: Accelerating scientific research with Gemini in the real world ↗Co-Scientist validation.
- arXiv: ArcticSwarm, deferring early consensus in multi-agent research ↗Sep 1, 2026.
- DeepMind: Co-Scientist published in Nature ↗May 19, 2026.
Watch next
- Find the arXiv perspective on trust, credit and incentives for agent research.
Sources
- arXiv: AI Agent Swarms as Researchersarxiv.org
- arXiv: A Society of Researchersarxiv.org
- The Decoder: DeepMind's AI Co-Scientist now runs lab equipmentthe-decoder.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 1 October 2026 at 21:47 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →