NeurIPS 2020 Grapples With Ethics, Bias, and Big Tech Influence
The world's largest AI research conference highlights tension between corporate power and ethical accountability, sparked by Timnit Gebru's firing and sponsor controversies. Researchers demand systemic change to address bias and shift control back to marginalized communities.
Organizers of NeurIPS award Best Paper honors to three research projects, including OpenAI's GPT-3 language model, while a significant boycott takes place as AI researchers refuse to review Google's papers. This protest follows the controversial firing of Google Ethical AI co-lead Timnit Gebru, an action that employees denounce as unprecedented research censorship. The situation highlights the immense influence Google holds in the AI space, as the company currently publishes more AI research than any other institution.
Ethical tensions dominate the conference workshops, where participants explore GPT-3's anti-Muslim bias and the use of AI for surveillance against minority groups. These discussions gain urgency following reports that platinum sponsor Huawei develops AI tools used to track Uighur Muslims in China. In response to these concerns, NeurIPS announces the formation of a new sponsorship committee to establish clear policies for vetting and accepting corporate sponsors.
The Resistance AI workshop challenges the industry's current trajectory by comparing Big Tech's research funding tactics to those historically used by Big Tobacco. Organizers express frustration with the limitations of "AI for good" initiatives, warning that such framing often serves as ethics-washing. Instead, they champion an approach to artificial intelligence that actively returns power to marginalized people and rejects systems of oppression.