Why I didn't sign the "We Must Act Now" statement (yet)
Noah Smith
· Noahpinion
· July 15, 2026 at 08:19
· ⏱ 9 min read
| Read on Substack ↗
Summary
Noah Smith declines to sign a vague 'We Must Act Now' statement on AI's economic transformation because its call to 'steer AI' is both impractical and based on flawed assumptions about job displacement. He argues that current data shows no macro evidence of AI destroying jobs, that technologists have a poor track record predicting labor effects, and that any attempt to redirect AI innovation would likely be ineffective or harmful. For markets, the piece reinforces a skeptical view of policy interventions in AI development and suggests the economic disruption narrative may be overblown.
•The statement argues AI may become 'radically more powerful' over the next 10 years and could cause 'large-scale job displacement' or 'major gains in living standards', but offers no specific policies.
•Smith believes signing the vague statement would implicitly endorse future unknown policy proposals, particularly the idea of 'steering' AI to complement humans, which he views as misguided.
•Daron Acemoglu, a signatory, helped revise the statement and champions the 'steering' idea from his book 'Power and Progress'.
•Smith points to macro data showing employment rates for age 20-24 and 25-54 are unchanged since ChatGPT's launch, and a study finding AI-adopting companies hire more workers than peers.
•He cites Geoffrey Hinton's failed prediction that AI would eliminate radiologists, noting instead a boom in radiologist hiring and salaries, as evidence technologists cannot forecast labor effects.
•Smith argues that historical heavy-handed interventions in technology (e.g., Mao's 'backyard production') caused harm, and that 'inaction' may be the optimal economic approach to AI.