NeurIPS 2020 Highlights Tension Between Corporate Power and AI Ethics
The NeurIPS 2020 conference showcases award-winning AI models while researchers protest Google's firing of Ethical AI co-lead Timnit Gebru. Workshops throughout the event highlight deep concerns over corporate censorship, anti-Muslim bias, and the monopolistic control of Big Tech.
NeurIPS 2020 opens with organizers awarding Best Paper honors to OpenAI's GPT-3, but the week is immediately overshadowed by a researcher boycott of Google's AI papers. This protest follows the firing of Ethical AI team co-lead Timnit Gebru, an act Googlers describe as unprecedented research censorship from the world's top publisher of AI research. The conflict sets a contentious tone for the entire conference.
Tension between corporate interests and human rights dominates various workshops, particularly a session that explores GPT-3's anti-Muslim bias and the use of AI to surveil oppressed groups. These discussions gain urgency following reports that platinum sponsor Huawei helps power a "Uighur alarm" for Chinese authorities, prompting NeurIPS organizers to form a new committee to evaluate and vet future sponsors.
Later in the week, Microsoft executive Chris Bishop defends Big Tech's infrastructure monopoly by pointing to the accessibility of cloud computing. However, the Resistance AI workshop pushes back by comparing corporate research funding tactics to those of Big Tobacco, arguing that the industry must move beyond performative "AI for good" initiatives to actively dismantle oppressive systems and return power to marginalized communities.