AI Economics

Paper · AI R&D · Growth & productivity · Economic measurement · Automation & tasks

Can AI agents conduct open-ended AI research?

Peter Kirgis, Sayash Kapoor, Andrew Schwartz, Stephan Rabanser, David Africa, Konstantinos Voudouris, Viet Nguyen, Toby Pilditch, Magda Dubois, Harry Coppock, Cozmin Ududec, Nitya Nadgir, Matilda Orona, Tilman Bayer, Derrick Chan-Sew, Yue Ling, Abhishek Shetty, Helen Toner, Gillian Hadfield, Seth Lazar, Steve Newman, Shoshannah Tekofsky, Rishi Bommasani, and Arvind Narayanan

· Revised 7 August 2026

Diagram of the long-running agent loop used in the CRUX 2 research evaluation.
Source visual: Kirgis et al.

Summary

In two shadow evaluations based on unpublished AI papers, research agents completed substantial engineering work but made no substantial research progress. The agents struggled with judging whether results were publishable, generating creative directions, backtracking after failure, managing resources, and staying aligned with the research goal.

Why it matters

The case studies provide direct negative evidence against treating strong coding performance or benchmark success as proof that present agents can autonomously advance open-ended AI research.

What to keep in mind

The sample contains only two research projects, models and scaffolds are changing rapidly, and the original paper authors helped evaluate the outputs. The result does not imply that agents cannot accelerate narrower or more structured parts of research.

Sources