MIT says the paper should “retreat from public discourse” due to concerns about the “integrity” of a well-known paper on the impact of artificial intelligence on the productivity of the Institute of Materials Science.
The paper in question, “Artificial Intelligence, Scientific Discovery, Product Innovation,” was written by doctoral students in the university’s economics program. He argued that the introduction of AI tools to the non-large, but unified Institute of Materials Science revealed more materials and patent applications, but showed the sacrifices to reduce researchers’ job satisfaction.
MIT economists Daron Acemoglu (who recently won a Nobel Prize) and David Autor praised the paper last year, telling the Wall Street Journal that it was the “floor.” In a statement included in MIT’s announcement on Friday, Acemoglu and Autotor described the paper as “already known and discussed in the literature on AI and science, despite not being published in the judged journal.”
However, the two economists said they were “not confident in the source, reliability, or validity of the data, and the truth behind the research.”
According to the WSJ, computer scientists with experience in materials science approached Acemoglu and Autor with concern in January. They brought those concerns to MIT, leading to internal reviews.
According to MIT, student privacy laws do not allow us to disclose the results of that review, but the author of the paper states that it is “not MIT anymore.” Additionally, university presentations do not name students, but both pre-edition papers and initial press coverage identifies the author as Aidan Toner-Rodgers. (TechCrunch contacted Toner-Rodgers for comment.)
MIT also states that it requested that the paper be withdrawn from the Quarterly Journal of Economics. Apparently, only the authors of the paper can submit a request for withdrawal of Arxiv, but MIT says, “To date, the authors have not.”