MIT has now stepped in to call for the withdrawal of a paper that linked an AI tool to higher productivity in materials science labs. This study, which made waves for suggesting that an AI system boosted material discoveries and patents at a major lab—albeit at the cost of researchers’ job satisfaction—was once championed by notable economists such as Nobel laureate Daron Acemoglu and David Autor.
However, when a seasoned computer scientist working in materials science flagged issues with the data, doubts quickly emerged. MIT launched an internal review, but due to student privacy laws the findings weren’t made public. The paper’s author, a doctoral student identified as Aidan Toner-Rodgers, is no longer affiliated with the institute.
In light of the growing concerns, MIT has asked both The Quarterly Journal of Economics and the preprint site arXiv to remove the paper. Still, because only the author can initiate a withdrawal on arXiv, the paper remains online. For anyone grappling with the challenges of ensuring data integrity in research, this episode underscores the crucial need for transparency and robust validation.
