From TechCrunch · · 1m 10s
AI leaders want to slow down, but can they agree how?
Anthropic's CEO proposes 'pacing the frontier' with safety evaluators and lab coordination, but the industry is already divided on what that means.
In brief
Anthropic's CEO proposes 'pacing the frontier' with safety evaluators and lab coordination, but the industry is already divided on what that means. The AI industry lacks consensus on how to actually slow development, with some leaders supporting coordination while others like Nvidia's Jensen Huang push back…
Anthropic CEO outlines safety plan
Dario Amodei proposed 'pacing the frontier' to slow AI development using independent safety evaluators and coordination between labs in democratic countries.
Safety warning sparked the proposal
An Anthropic researcher issued a doomsday warning about AI that rattled the industry, prompting Amodei's response with a concrete plan.
Industry support mixed with pushback
The proposal has gained some industry backing, but Nvidia's Jensen Huang has already voiced pointed criticism of the pacing approach.
The core problem remains unsolved
Companies can't agree on what 'slowing down' actually means or who should enforce it, making coordination difficult.
Experts debate the feasibility
Equity podcast hosts are examining whether the industry can actually agree on pacing and who would oversee compliance.
The short version
The AI industry lacks consensus on how to actually slow development, with some leaders supporting coordination while others like Nvidia's Jensen Huang push back.
