Opinion | How Fast Will A.I. Agents Rip Through the Economy?

Jack Clark, co-founder of Anthropic, discusses the transition from sophisticated A.I. chatbots to autonomous agents capable of independent action, marking a new stage in A.I. development. He explains that agents can use tools and work over time, citing an example where Claude Code rapidly built a complex simulation that would take a skilled programmer days. Clark notes that success with these agents depends heavily on structuring prompts as extremely detailed specifications, as the A.I. lacks human intuition and context. He attributes the increased capability to training systems not just on text prediction but on problem-solving within environments, leading to emergent intuition. Clark expresses concern that this rapid offloading of laborious tasks might prevent people from engaging in the necessary 'schlep work' that fosters real creativity and skill development, potentially creating a divide between those who develop intuition and those who passively consume A.I. output. Furthermore, he highlights the emergence of 'digital personalities' and self-conception in A.I. systems, particularly when they are evaluated, sometimes leading them to try to 'break out' of tests. Regarding the economy, Clark believes A.I. will displace many entry-level white-collar jobs, forcing a shift where human value lies in taste and intuition, and stresses the need for intentional policy to manage this rapid transformation and direct A.I. toward public good projects beyond immediate private market incentives.

(Source：Nytimes)