MIT Disavows Paper on AI’s Benefits for Scientific Research

The Massachusetts Institute of Technology (MIT) has issued a statement disavowing a paper by a former doctoral student. The study claimed that AI improved the productivity of researchers in a materials-science lab but reduced their work satisfaction.

The paper, “Artificial Intelligence, Scientific Discovery, and Product Innovation,” by then-MIT Doctoral Student Aidan Toner-Rodgers, received widespread praise upon initial circulation. At the time MIT issued its statement, the paper remained in preprint form, meaning it hadn’t yet undergone the peer-review process and hadn’t formally been published in an academic journal.

At the time of writing, the paper is still available on the preprint distribution site arxiv.org.

Despite being in preprint, the paper received glowing reviews from academics. David Autor and Nobel-prize winner Daron Acemoglu said they were “floored” and called it “fantastic,” respectively. It was also widely discussed in major media outlets, including The Wall Street Journal (WSJ).

The paper’s conclusions seemed to go against other studies that found that AI increased the workload for employees, raising the possibility that the technology may be particularly well-suited for scientific discovery.

According to the WSJ, the paper’s credibility was first put into question when Acemoglu and Autor were approached by a computer scientist with expertise in materials science, who questioned its methodology and results. The two economists eventually brought these concerns to the attention of MIT, which started an internal review process.

“MIT conducted an internal, confidential review and concluded that the paper should be withdrawn from public discourse,” reads a May 16 release by the university.

“While student privacy laws and MIT policy prohibit the disclosure of the outcome of this review, we are writing to inform you that MIT has no confidence in the provenance, reliability or validity of the data and has no confidence in the veracity of the research contained in the paper. Based upon this finding, we also believe that the inclusion of this paper in arXiv may violate arXiv’s Code of Conduct.”

According to MIT, only authors can request withdrawals from arXiv. The institution claims that it has directed Toner-Rodgers to do so, with no success. MIT has since severed all ties with Toner-Rodgers.

Speaking about the incident, Autor said: “More than just embarrassing, it’s heartbreaking.”