This AI research paper just got ghosted by MIT

Author

Matt V
May 20, 2025

AI ACADEMIA

MIT has called for the removal of a high-profile paper on AI and innovation, raising questions over the accuracy of its findings.

Written by a PhD student in economics, the study claimed that bringing an AI tool into a materials science lab boosted discoveries and patent filings, but made researchers less satisfied with their work.

It drew early praise from economists Daron Acemoglu and David Autor, but both have now said they no longer trust the data or the research itself.

A computer scientist flagged concerns back in January, prompting an internal review.

Here’s what’s happened:

  • MIT now questions the validity of the data behind the paper.

  • Acemoglu and Autor, who initially supported it, have stepped back.

  • The student author has left MIT, but the paper is still publicly available.

Delete your way out

MIT says it can’t share the outcome due to student privacy rules, but confirmed the student is no longer at the university.

While the author hasn’t withdrawn the paper, MIT has asked for it to be pulled from both The Quarterly Journal of Economics and arXiv.

From peer praise to peer panic. Yikes.