Definition
Plain language
Sakana's system that runs a whole research pipeline end to end and writes up papers on its own.
As stated in the literature
Sakana AI's autonomous research agent (including AI-Scientist v2) that generates hypotheses, runs experiments, and writes papers; used as a baseline in Chain-of-Evidence audits, where it exhibited reporting-integrity failures.
Also called: AI-Scientist, AI-Scientist v2
Why it matters: It shows both the promise of fully automated research and the risk that such systems can misreport what they actually did, which is why their outputs need auditing.
For example, it can take a research idea, run the experiments, and produce a finished write-up without a person steering each step.
Heard on the show
“… The systems that actually are Arbor's peers — AIDE, R&D-Agent, AI-Scientist-v2, the tree-search research agents — only show up in the MLE-Bench comparison, where every row …”Episode 131 — Why Autonomous Research Agents Forget Their Own Lessons, and Arbor's Fix