MIT has publicly disavowed a high-profile research paper authored by a doctoral student that claimed significant productivity gains resulting from artificial intelligence implementation in a materials science laboratory. Citing concerns over the integrity and reliability of the presented data, university officials recommended withdrawing the study from circulation.
The doctoral candidate, affiliated with MIT’s economics program, authored the paper titled “Artificial Intelligence, Scientific Discovery, and Product Innovation.” The research received notable attention for its conclusion that integrating AI tools significantly boosted the discovery of new materials and led to increased patent filings. However, the study also noted negatively impacted researcher satisfaction levels.
Initially praised by prominent MIT economists, including Nobel Laureate Daron Acemoglu and his colleague David Autor, the paper had garnered considerable academic interest. Previously, Autor publicly expressed astonishment at the findings, underscoring the seemingly groundbreaking nature of the claims. Nonetheless, recent concerns have led both Acemoglu and Autor to reverse their support, jointly stating that they now lack “confidence in the provenance, reliability or validity of the data and in the veracity of the research.”
The doubts regarding the research were brought forward in January by a computer scientist with expertise in materials science, prompting MIT to launch an internal inquiry into the allegations. While the institution noted limitations on disclosure due to student privacy laws, MIT confirmed that the author of the paper is no longer associated with the university.
Reports state the paper’s author as Aidan Toner-Rodgers, a former doctoral student previously identified through preprint publications and media coverage. MIT officials requested the paper’s withdrawal from The Quarterly Journal of Economics, where it was originally submitted, and from the online preprint platform arXiv. However, the university clarified that such a request must originate from the author directly, and that at the time of their announcement, no withdrawal from arXiv had yet been initiated by Toner-Rodgers.
