Google says it's committed to ethical AI research. Its ethical AI team isn't so sure.
ai-ethicsgoogletimnit-gebruresearch-culture
Abstraction: Google ethical AI team dysfunction after Timnit Gebru firing
Key points:
- Six months after Timnit Gebru was fired over a paper on large language model harms, Google's 10-person ethical AI group was leaderless and self-managing ad-hoc
- Meg Mitchell (co-founder of the team) was fired in February 2021; Samy Bengio resigned in April 2021 and joined Apple
- Marian Croak appointed head of new Responsible AI department, but researchers felt she made empty promises and lacked day-to-day involvement
- Academic backlash was severe: thousands signed a petition, FAccT suspended Google's sponsorship, Luke Stark rejected a $60,000 Google grant, Black in AI and Queer in AI rejected Google funding
- Google pledged to double ethical AI research staff to 200, but team members doubted institutional reform was possible
Connections: Google · Timnit Gebru · Jeff Dean · AI Ethics · Algorithmic Fairness