DeepMind Co-Founder Legg Joins Safety-First Camp in AI Debate
Shane Legg, co-founder of Google DeepMind, has joined the growing camp advocating for prioritizing AI safety over rapid advancements in capabilities. In an interview with the Financial Times, Legg emphasized that AI progress must not outpace safety measures, saying 'we can't let capabilities get ahead of safety.' This stance comes amidst a heated debate in Silicon Valley about whether to slow down the release of new AI models due to concerns over potential risks.
Legg's comments followed an essay by Anthropic CEO Dario Amodei titled 'We Must Pace the Frontier,' which proposed slowing down AI development to focus on safety. While Legg described Amodei's idea as 'an interesting directionally' and 'worth considering,' he noted that it would require a deeper exploration of its practical implications.
Legg also announced the launch of the DeepMind Institute, aimed at examining the consequences of developing artificial general intelligence (AGI), which refers to systems with cognitive capabilities rivaling those of humans. Despite recent pronouncements by executives at OpenAI and Nvidia suggesting that AGI is near, Legg expressed caution, stating he remains 'comfortable' with his long-held prediction that there's a 50% chance of minimal AGI by 2028.