Member-only story
Noted A.I. Ethicist Timnit Gebru Let Go From Google Following Tense Email Exchange
Gebru is known for influential research about bias in facial recognition
Timnit Gebru, a pioneering researcher on algorithmic bias, said Wednesday night that she had been abruptly let go by Google, where she was technical co-lead of the company’s Ethical Artificial Intelligence Team, after she had privately threatened to resign.
Gebru is known for her co-authorship with Joy Buolamwini of an influential 2018 paper on bias in facial recognition software, among other work. The study found that three leading facial recognition systems were far more likely to misidentify women and people of color than white men. The findings helped to fuel a backlash against facial recognition that has led some major companies and jurisdictions to stop developing or using the technology. OneZero’s Dave Gershgorn wrote in June about the study’s profound impact.
Gebru also co-founded the group Black in A.I., which works on both diversity in the A.I. field and bias in the data sets it relies on.
Gebru told OneZero she was let go via email by Megan Kacholia, a vice president of engineering at Google Research, who in turn reports to the company’s famous A.I. lead, Jeff Dean. Gebru said the move stemmed from an email Gebru had sent to a pair of internal Google employee groups, which upset Google executives. She said she could not immediately provide the contents of that email because she had been cut off from her corporate accounts.
According to Gebru, the email led to an exchange with managers in which Gebru apparently laid out conditions under which she’d continue working for the company. Otherwise, Gebru told them, she would work with her…
