Google loses DeepMind researchers over military AI use

Two researchers from Google DeepMind have resigned in dissatisfaction over the collaboration between Google and the U.S. Department of Defense. Sofia Liguori and Andreas Kirsch have criticized the possibility that the military may use Google's AI models for purposes they cannot control. Liguori stated that she feels ashamed to work at Google, while Kirsch emphasizes that Google has no veto power over how the AI is deployed by the Pentagon.

The reason for their departure is a contract that Google signed with the Pentagon in the spring. This contract allows the military to use Google's AI technology in classified environments for various government purposes. Although Google claims that the AI cannot be used for mass surveillance or autonomous weapons without human oversight, the researchers have serious doubts about the actual impact of these agreements, given the lack of control and transparency.

The criticism from both researchers stems not only from their personal experiences but also from a broader dissatisfaction within Google. Employees are concerned about their influence and the ability to voice criticism. Liguori notes that there is a culture of fear and that many colleagues feel they are being monitored. Google has since stated that employees can safely express their concerns, although in practice, employees refrain from criticism due to strict contractual clauses.

The issue of whether and how human oversight is applied to AI applications in military environments is of great importance. The Netherlands has taken the initiative in 2023 for a global summit on the responsible use of AI by militaries, with 60 countries signing that AI must remain under human control. However, it remains unclear how this oversight will actually be realized, especially given the self-regulating nature of contracts like Google's with the Pentagon.

Read the full article from AI Insider.