The trade-off is a fiction, and the fiction is winning
A coordinated slowdown is worth pursuing even at the cost of advantage, because uncoordinated speed is the greater risk.
The choice that isn't there
The safety argument and the speed argument are not two camps. They are one argument made twice, and the sources making it say so plainly 1.
What divides them is not whether care matters. It is whether care can be scheduled — a different argument, and a harder one 2.
Pacing is not a brake. It is a schedule: what gets released, when, and what evidence has to exist first.
It was renamed
That argument is losing, and not on the merits. It is losing because it was renamed.
Call it a hoax and you never have to answer it 4. The denying side is not short of confidence.
What the denials argue
The move works by making care sound like delay. Every attempt to answer the charge arrives looking like hesitation, which is the cost the charge was designed to impose 5.
Read the denials closely, though, and most are not arguments against care. They are arguments against scheduling care — a distinction the denials themselves make 6.
Read the primaries yourself
So the false choice is doing real work. Whoever names the choice decides which side has to look reasonable.
The trade-off between safety and speed is a fiction. The fiction is winning. The only way to see that is to read the primaries 1 4.
Sources
-
The idea of pausing or slowing AI has been floated [as far back as 2023](https://futureoflife.org/open-letter/pause-giant-ai-experiments/), and I think it made little sense back then. The question was always: _what would you do with the extra time_? The AI models of those days were not powerful enough to act as agents in the world in any coherent way, and were not capable of significant deception, manipulation, cheating, or cyberattacks. Slowing down in order to address their alignment risks felt like trying to study the psychology of humans by performing experiments on bacteria.
-
But over the last few months, I have become convinced that fully addressing the risks requires even more prudence — not just investing in risk prevention, but pacing the rate of capabilities advancement so that risk prevention has time to keep up. **We must slow the pace at which we improve the capabilities of AI models.** **Progress will still seem fast, and we must make wise use of the time we gain.** Two things have convinced me.
-
Pacing within democracies will be limited by the lead that US companies have over authoritarian regimes, chiefly the Chinese Communist Party. If we slow down by more than this amount, then (unpaced) CCP-associated projects will pull ahead, creating significant national security risk. I agree with [Secretary Bessent](https://www.bloomberg.com/news/articles/2026-09-09/bessent-warns-nothing-would-matter-if-china-wins-the-ai-race) that a Chinese lead in AI would pose grave danger for the United States and the world.
-
Um I I think that that um regulation uh of the technology uh needs needs to think about the unintended consequences of regulation. Um um we we need to think about think about uh AI not in the con context of some kind of a foot race in a 100 meter dash. Um, the narrative that whoever gets to the 100 meter dash, whoever wins that wins it forever is nonsense. The idea that >> And here you're talking about the US and China. >> That's right. It's complete nonsense. All of those stories being made up is scaring Washington DC. We're >> And do you think it's scaring the White House?
-
Uh we have to recognize that China has some of the largest population of science and math students in the world. They're if they manufacture anything that's incredible. They manufacture math students incredibly. And so so they they uh they manufacture more AI researchers than the rest of the world combined. So winning. >> So the fact that China is going to be extraordinary at this is a foregone conclusion. And the idea that we would hold China back is illconceived and and uh it's not going to happen. Um nor are they going to hold the United States back.
-
Thank you. I don't think you'll find much disagreement with that proposition. But to have American workers do those jobs, we need to train them. And you all in some sense, because you're all teachers, you're all professors are engaged in that enterprise. Mr. Amodei, I don't know whether you can be called still a professor, but probably not. I was never a Professor <laugh>, but we need to train workers to do these jobs, and for those who want to pause, and some of the experts have written that we should pause AI development I don't think it's gonna happen.