Google to restructure DeepMind, moving 90-person AI ethics team

Perceive Beating AI News: According to The Wall Street Journal, Google plans to move about 90 members of its "AI Ethics" team from Google DeepMind to the company's global affairs division, with the restructuring expected to take effect in early September. This division is responsible for Google's overall lobbying and public policy efforts, and oversees the group-level Trust & Safety team.

This adjustment is part of Google's further integration of DeepMind. Previously, DeepMind's co-founder and CEO Demis Hassabis transitioned to the role of chairman, focusing on external engagements and AI's broader societal impacts; his successor Koray Kavukcuoglu will lead DeepMind as a Google Senior Vice President rather than CEO. Teams such as Human Resources and Policy will also be transferred from DeepMind to Google's central division.

The AI Ethics team's responsibilities include testing Google's AI models for potential chemical, biological, radiological, and nuclear risks, as well as researching user interactions with AI and the psychological effects of chatbots. Some employees are concerned that after the move from DeepMind, the team's access and understanding of the Gemini model's development work may decrease, impacting their ability to conduct independent research and identify risks in new models.

Google responded by stating that consolidating the AI Ethics team's management can enhance its influence on model and product safety efforts. Some employees had applied to transfer to stay in DeepMind's research team but were rejected; management also acknowledged that the restructuring might lead some employees to join other AI labs. This adjustment comes as several senior researchers from DeepMind have been leaving the company.