Extinction risk is not the right first sentence

Per LessWrong contributor Michael Wilkinson, the AI safety field has been leading with extinction risk—the least effective message for moving people to act. In a randomized trial, existential threat framing scored just 14 out of 100 on willingness to take action. Concrete harms mobilize far more. A Pennsylvania town's petition against a data center collected over eighteen hundred signatures and killed the project; across the country, 800-plus community groups in 49 states are organizing against AI data centers. Community opposition stalled more than one hundred fifty-six billion dollars of construction in 2025 and blocked or delayed another one hundred thirty billion dollars in early 2026. Wilkinson, a tech founder, argues people dismiss abstract existential risk but respond powerfully to proximate harms—depleted water, rising energy bills, disrupted landscapes. He suggests the AI safety field should reframe around these concrete concerns to build lasting public pressure and regulations that survive industry lobbying.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton