AI Leaders Clash Over Development Pace and Safety
Dario Amodei calls for a global slowdown in AI advancement to prevent systemic risks, sparking a debate among tech leaders and policymakers.
Anthropic CEO Dario Amodei published a manifesto titled "We Must Pace the Frontier," urging a deliberate slowdown in artificial intelligence capabilities to allow risk prevention to keep pace. Amodei warns that recursive self-improvement and the potential for AI agents to take over the internet within six to 12 months pose significant perils. He proposes embedding evaluators within AI firms and coordinating with democratic leaders and authoritarian competitors, specifically China, to limit the pace of progress.
Other industry figures reject this cautious approach. Marc Andreessen argues in his "Techno-Optimist Manifesto" that decelerating technology is evil, asserting that AI serves as a universal problem solver. Bill Gates maintains that global deceleration is unlikely because of economic and geopolitical incentives, though he expressed openness to a credible plan for slowing advances.
In the political sphere, President Donald Trump entered a morally binding agreement with technology executives to implement AI audits and third-party evaluations. While Trump suggests the arrangement will foster self-policing, the agreement lacks formal enforcement mechanisms. Simultaneously, Pope Leo XIV published the encyclical "Magnifica Humanitas," calling for a broader debate on human values rather than focusing solely on technical AI alignment.
