In a shocking revelation reported by American media, the economics thesis of a doctoral student at the Massachusetts Institute of Technology (MIT) was found to be fabricated. The school has since initiated discussions on how to improve the standards of graduate theses and is requiring students to prove that their research is not falsified.
At the end of 2024, a 27-year-old MIT doctoral student, Aidan Toner-Rodgers, caused a sensation with his thesis on the impact of artificial intelligence (AI) on the workplace, which even garnered attention from the US Congress. However, suspicions were raised as the data and arguments presented in the thesis were deemed too perfect to be true. Toner-Rodgers was subsequently dismissed from MIT in the spring of the following year.
The Wall Street Journal reported on Wednesday (December 3) that Toner-Rodgers’ former colleagues and mentors were shocked by his dishonesty, as his fraudulent conduct went beyond simply manipulating variables and involved fabricating the entire research project.
According to insiders cited by the Journal, following the scandal, MIT’s economics professors began discussing ways to enhance the standards of graduate theses, including stringent scrutiny of original data and requiring students to diligently prove the authenticity of their research.
So, what did Toner-Rodgers do exactly? First, he fabricated the entire research subject and data sources; second, when requested to withdraw data by the research subject, he falsified data usage agreements; third, he registered fake domain names and created fake emails/fake official websites; fourth, he fabricated all empirical results and statistical data; fifth, he falsified survey responses and statements from interviewed scientists.
As reported by the Journal, in 2024 Toner-Rodgers told others that he had an old classmate working at a large company involved in materials science, researching new material forms applicable to biomedical devices and other technologies. He hinted at possibly obtaining extensive AI experiment data mentioned by this friend.
During a lunchtime lecture on campus, Toner-Rodgers, in front of over 20 faculty and students, shared a case study involving 1,018 scientists at an American major enterprise’s material molecular laboratory utilizing AI tools. He claimed that scientists only needed to set the desired compound characteristics, and the AI tools would generate formulations for their selection.
In November 2024, Toner-Rodgers presented his findings at a seminar hosted by the prestigious National Bureau of Economic Research, receiving enthusiastic feedback. The thesis was later cited in a congressional hearing that same month, and several weeks later it was highlighted in the scientific journal Nature. The final version was released in December on the global research sharing platform arXiv.
Charles Elkan, a computer scientist at the University of California, San Diego, who previously held senior positions at Goldman Sachs and Amazon leading machine learning teams, came across the thesis in an email and found something amiss.
Elkan, having knowledge of American materials companies, noted that only 3M and Corning had employed numerous scientists. He questioned why these companies would keep AI-developed materials so secretive if they had indeed created them, finding it nonsensical.
On January 5, 2025, Elkan sent an email to Toner-Rodgers’ two mentors raising concerns about the thesis’s academic integrity.
Corning lodged a complaint with the World Intellectual Property Organization on January 31 against the registrant of the domain corningresearch.com, asserting that the domain could mislead the public into associating it with Corning. The written decision by the WIPO revealed Toner-Rodgers as the domain’s registrant.
In May of this year, MIT released a statement disavowing Toner-Rodgers’ thesis, expressing no confidence in the data sources, reliability, or validity, as well as the authenticity of the research contained in the thesis. Although the statement did not directly name Toner-Rodgers, it mentioned that the “author has left MIT.”
Both Corning and 3M informed the Journal that they did not conduct the experiments as described by Toner-Rodgers nor share data with him.