The A.I. Slowdown Debate Goes Global

ImageAndrew here.The weekend was one long freakout over the existential risk of A.I., set off not by critics, but by the people building it.Anthropic, OpenAI, SpaceX, Alphabet and Microsoft are all calling for government safeguards.

Is this sincere, or just a bid for regulatory capture? Even a real slowdown in A.I.models’ training won’t change the math on data centers and chips, at least in the near term.

The supply chain is too constrained.(Notably, Jensen Huang of Nvidia is not saying he’s scared.)As for the cynics: When Elon Musk, Sam Altman and others started OpenAI, they spoke openly about A.I.’s risks.

So have Dario Amodei and Demis Hassabis.I’ve been talking to them — on the record — about this very issue for a decade.

Could tighter regulation make closed models, like those from Anthropic and OpenAI, more valuable? Sure.But that’s not where these conversations started.In the immediate term, here’s what A.I.

leaders clearly want: permission to talk to one another.Is that collusion or an antitrust problem? Maybe.

More below.Also, we have an exclusive: Ron Conway and Jay Carney are teaming up on this very issue.ImageTo pace or not to paceIt’s rare to see Dario Amodei of Anthropic and Sam Altman of OpenAI agree on much.

So the world was struck when the leaders of those two artificial intelligence giants called for a slowdown — “pacing,” in Amodei’s words — in developing bleeding-edge models.We are having trouble retrieving the article content.Please enable JavaScript in your browser settings.Thank you for your patience while we verify access.If you are in Reader mode please exit and log into your Times account, or subscribe for all of The Times.Thank you for your patience while we verify access.Already a subscriber? Log in.Want all of The Times? Subscribe....