Shane Legg, co-founder of Google DeepMind, recently emphasized the need to prioritize safety as AI capabilities evolve rapidly. He noted that the proposal by Anthropic CEO Dario Amodei to pace AI development is worth considering, though practical implementation details remain to be addressed.
Legg officially launched the DeepMind Institute on Wednesday. The organization aims to analyze the implications of artificial general intelligence, which the company defines as systems possessing cognitive capabilities comparable to the human brain.
Regarding the timeline for AGI, Legg maintains his projection of a 50% probability of achieving minimal AGI by 2028. He stated that it is currently premature to declare that AGI has already arrived, despite recent claims from other industry executives.
Meta CEO Mark Zuckerberg also weighed in on the safety debate, arguing that trust and alignment are becoming critical differentiators for AI models. He noted that Meta previously delayed the release of its Muse AI to focus on safety and security measures.
Zuckerberg highlighted that AI labs have strong incentives to prioritize safety due to potential liability risks. He suggested that industry best practices should include engaging independent evaluators and dedicating compute resources to user-facing tasks.
