Agentic AI Scientists Are Not Built For Autonomous Scientific Discovery

Summary

Position paper naming four structural reasons agentic AI scientists work as co-scientists but not as autonomous discoverers: McNamara-fallacy problem selection, training corpora that omit tacit procedural and failure knowledge of lab practice, preference optimisation that compresses output diversity toward consensus, and benchmarks that measure single-turn prediction without physical-experiment feedback. Proposes simulations as training verifiers, persistent world models, and a centralised preregistration repository for AI-generated hypotheses — a governance idea directly transferable to IS.

Contribution

Queued — not yet read.

Method

Queued — not yet read.

Relevance to RISE

Surfaced by the 2026-08-29 literature sweep as a medium-priority candidate for the RISE knowledge base (angle: autonomous AI scientists and their evaluation). Verified against: arXiv abstract page opened 2026-08-29 (title, 5 authors, submitted 9 May 2026, full abstract read)

Critique / open questions

Queued — not yet assessed.

Key quotes

A growing body of work pursues AI scientists capable of end-to-end autonomous scientific discovery. (abstract)

This position paper argues that although they already function as co-scientists, agentic AI scientists are not built for autonomous scientific discovery. (abstract)