AI tech bros are self-serving and slimy. But we must heed them anyway and pause development

Chris Gay is a contributing columnist for The Globe and Mail. He is a former Wall Street Journal staffer and writes the newsletter Figure at Center.

If you’re starting to find the artificial intelligence arms race confusing, that’s because it is. Anyone who’s watched Silicon Valley knows that Big Tech is not inclined toward modesty and self-restraint and isn’t exactly partial to guardrails. But AI’s biggest players are now calling for precisely that. We should heed them.

Alert readers also know that under normal conditions business-friendly lawmakers and right-wing think tanks tend to howl about getting government out of the way. Now, even wholly owned corporate Republicans who get paid to hate Big Government are demanding intervention, as are a large majority of U.S. voters.

AI authority and economic historian Donald Trump appears to be the loudest contrarian. He says tighter safety measures would mean forfeiting victory in the AI race. (You’re not supposed to remember that he signed an executive order in June calling for federal oversight.)

“The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!” Mr. Trump posted on Truth Social on Monday.

Opinion: AI’s Hiroshima moment may be near

While it’s hard to argue with all those capital letters, people who actually know what they’re talking about say otherwise. AI researcher Jacob Coxon, a 27-year-old Briton and now AI’s very own Edward Snowden, resigned loudly from Anthropic, parent of Claude, last week. He’d spent three years working in pretraining (feeding models gargantuan volumes of data) that included time at OpenAI.

“Neither company is acting responsibly,” Mr. Coxon said on X, but instead “are racing straight to self-improving superintelligence and gambling with our lives. … The people building AI earnestly believe that it could kill us all by the end of the decade.”

This seems important, which is maybe why Anthropic chief executive officer Dario Amodei posted an essay days later suggesting that the industry press pause until safety measures catch up. OpenAI CEO Sam Altman (ChatGPT), former Google AI chief Demis Hassabis and SpaceX CEO Elon Musk (SpaceXAI) quickly signed on.

Skeptics smell a ruse. “Stop pretending the motivation to slow down is purely altruistic,” posted Trump tech adviser David Sacks. Nvidia CEO Jensen Huang says frontier labs are simply clearing a path for their own cybersecurity products. “What better way to create demand than to create a problem?” he asks.

Trump is playing the America First card. “We’re leading China in AI,” he told reporters in Ireland on Sunday. “We’re the most sophisticated country in the world, and frankly I want to ⁠keep it that way because whoever wins AI, wins.”

AI firms’ calls for co-ordinated slowdown amounts to ‘cartel’ behaviour, Cohere CEO says

Smarter people than Trump say “race” is not the right framing here, but leave that aside. What would winning actually mean? How will we know the final score when we see it? Or is the fact that no one can answer that question grounds for a full-throttle gain-of-function throwdown, damn the consequences? True, the West needs to prepare for Chinese cybersabotage, but how would out-of-control AI help here?

Mere mortals are not equipped to answer such questions, but we can keep in mind two rules of thumb.

First, those tech bros. It’s best to weigh the sincerity of alpha-male-wannabe billionaires the way you’d size up hitchhikers in orange jumpsuits. Are these folks really concerned with unpalatable social consequences, or is their call for restraint just a high-stakes exercise in game theory? Maybe, to follow Mr. Sacks’s logic, they’re trying to concoct a kind of prisoner’s dilemma in which A calculates that B and C will stand down while A quietly outflanks them.

Second, ditch the trite analogies to previous disruptive technologies and Luddite reactionaries, unless you know something the experts don’t. True, movable type put scribes out of work and the moving pictures hollowed out vaudeville, but no one called them extinction events.

Editorial Board: The shocking and predictable threat posed by AI

Mr. Altman said three years ago that AI could bring “significant harm to the world” and proposed a government department to oversee it. Mr. Amodei said last year that there was a 1-in-4 chance that AI would lead to catastrophe, though he has since walked back the number. Biosecurity experts warn of chatbots helping to weaponize pathogens. Scientists already use AI to create new viruses.

Self-serving false equivalence is dangerous. AI is just the latest technology the way Little Boy was just the latest ordnance.

Whatever AI royalty’s motives for tapping the brakes, it’s probably best to take them at face value and heed the sentiments of Arizona Senator Ruben Gallego: “This is more akin to Dr. Frankenstein telling us, ‘The monster is loose, help me stop the monster.’”