Amodei CEO Proposes Slowing Down AI Development

Dario Amodei, CEO of Anthropic, has argued that the pace of improvement in artificial intelligence (AI) models should be slowed down to secure time for safety research. Elon Musk and Sam Altman have also agreed with this sentiment. Amodei emphasized the need to secure more time between significant enhancements in AI capabilities, stating that researchers and external evaluators should assess the risks of the systems before moving on to the next stage. He pointed out risks such as loss of control over advanced AI systems, misuse of cyber and biological attacks, and impacts on employment and the economy. Amodei believes that by slowing down and securing 1-2 years for alignment research, the risk of serious issues can be reduced. Anthropic plans to provide external evaluators with employee-level access to directly verify safety measures. External evaluators will be provided with office space, access badges, and company laptops to inspect the AI training systems and safety controls. Amodei stressed the importance of ensuring that the pace of American companies does not fall behind China's AI projects, mentioning the need for a regulatory framework that demands transparency, independent audits, and ongoing evaluations.