Amid the restructuring developments, Google has reportedly decided to move a team that studies the risks of AI out of DeepMind, the company’s primary AI research branch. According to The Wall Street Journal, this move is a part of major reorganisation and comes after Demis Hassabis announced that he is stepping down as CEO of DeepMind and is moving into a chairman role.
AI Safety Team To Move To Google’s Global Affairs Division
According to WSJ, the roughly 90-member ‘AI responsibility’ team will become part of Google’s global affairs organisation from the beginning of September, as per an internal email. Moreover, this shift is also reportedly raising concerns among some of the executives who fear it may make it harder for the team to carry out independent AI research.
The AI responsibility team works on several areas related to the potential risks of advanced AI systems. Its work includes testing AI models for chemical, biological, radiological and nuclear risks. This team also studies how people interact with AI systems and examines the psychological effects that chatbots can have on users.
Helen King, a Google DeepMind vice president who leads the responsibility team, allegedly explained the change in an internal email. “GDM is moving into a more typical PA setup with central teams in central Google,” King wrote. PA, or product area, is a term Google uses for major divisions such as Google Search.
Other DeepMind functions, including HR and policy teams, are also reportedly being moved into central Google as part of the restructuring. The team will continue to have access to DeepMind and will retain its existing computing resources and headcount, according to King. “Demis still cares a lot about Responsible AI and is still planning to be involved,” King wrote.
The restructuring also comes after several prominent researchers have left DeepMind in recent months. Earlier this month, Jeff Dean, a long-time figure in Google’s AI efforts, left the company along with three other senior DeepMind researchers to start a new company. The AI responsibility team’s move is expected to take effect at the start of September.
