Independent researcher
Shiro
Takagi
Building AI that conducts research — and rethinking how science itself should work.
Can a machine do science?
“I build AI that conducts academic research. And because peer review, papers, and universities are historical accidents rather than optimal designs, I am equally drawn to the other question: what new forms should science take when AI becomes a researcher? The design space of science has barely been explored — I want to create the new games of knowledge-making, not just play the old one.”
About me →Research
Publications and workshop papers on autonomous research agents, learning dynamics, and collective models of science.
- 01
- 02
- 03
- 04
Experiments
Open-source prototypes probing the infrastructure of autonomous science: what happens when AI runs hundreds of experiments per night?
-
01
Atlas of Judgment2026.09
What is actually inside a peer review? Every public ICLR review from 2018 to 2026 — about 200,000 reviews — read by an LLM pipeline into ...
-
02
Claim Prediction Market2026.04
Can prediction markets identify novel research claims? 5 LLM predictors evaluate 35 ARA-generated claims. Market ensemble achieves Brier ...
-
03
Research Anomaly Detector2026.04
Structural pathology detection for AI-generated research. 7 detectors find 47 anomalies in ARA output including confidence inflation (0.9...
Blog
Essays on autonomous research, metascience, and what research is for — plus public Lab Notes from work at Unktok.
-
UN
Atlas of Judgment: nine years of peer review, read one unit of reasoning at a time2026.09.01
Two reviewers write 5. One means the idea is not new; the other means the experiments do not support the claim. A score is a judgment wit...
-
UN
Unktok Reviewer: what happens when careful reading stops being scarce2026.08.19
Peer review rations the one resource it runs on — hours of close reading — and then compresses what it spends into a verdict. Our reviewe...
- 01 2026.06.12
- 02 2026.05.29
- 03 2026.05.29
- 04 2026.04.23