There are legitimate concerns about how quickly businesses are adopting frontier AI models and how rapidly their capabilities are growing.
I have pointed out on several occasions over the last year that the new models being released are coming "Too fast, too soon." But that doesn't mean calling a halt to the development of AI or putting restrictions on the technology that would force companies to slam the brakes on progress.
While the underlying arguments for a slowdown are based on facts or reasonable speculation, the anti-AI hysteria is all social media-generated. That much of it originates in China, and it doesn't seem to bother people like Sen. Bernie Sanders (I-V.T.), who seem perfectly willing to help Beijing get what it can't under ordinary circumstances: to leap ahead of the U.S. in this "next big thing."
The Free Press's Tyler Cowen offers a somewhat realistic scenario on what might happen over the next few years if we give in to our fears and listen to the Luddites and anti-tech crowd about AI. He writes from the perspective of offering a "post-mortem" on what went wrong.
"That fear was not based on peer-reviewed scientific research showing a high chance of doom, nor was doom indicated in any market prices of the time, including measures of risk," Cowen writes. Indeed, the "AI is going to kill us all" crowd isn't even talking about current AI models and how they're performing. They're worried about future models that employ artificial general intelligence, or AGI.
Halting the development of current models just doesn't make any sense. While safely developing new models, AI creators and engineers learn about the models' capabilities and how they can exploit weaknesses in other systems. Halting development will only slow the process of protecting ourselves.
There is a real danger that the current hysteria over data centers and advanced AI will result in crushing regulations that make AI all but useless or too expensive for companies to adapt. Those regulations won't keep us any safer, but they will allow China an open field in the race to scale up AI.
What would be the real-world result of these new AI "safety" regulations?
No new frontier AI models are released in the U.S. as companies sit and wait, and their revenue streams start to slacken. After all, a product that will not be upgraded is worth much less to users, especially big-spending corporate users.
But as the international negotiations proceed, it becomes clear that China is not so keen to share the Trumpian vision of safety through deals. It does want AI regulation, but for China that means American models should not be allowed to express what it considers anti-Chinese attitudes. Beijing is also suspicious of the progressive lean among the main American models (as displayed in a recent Washington Post study that found that ChatGPT and Claude expressed a liberal bias in 80 percent and 43 percent, respectively, of political answers). Threats to the Chinese Communist Party (CCP) are the main concern for Chinese leaders, and they do not worry very much about what Americans call “X risk”—the chance that superpowerful AI destroys all of humanity. They have a more pragmatic and less dramatic approach to the whole issue.
China would keep improving their models, secure in the knowledge that they won't feel the pressure from interest groups or the public about preventing the end of humanity. They will go on perfecting their models, adapting them to serve the state in a variety of fields.
After a few years Chinese AI companies have a clear lead, and their models are an order of magnitude better than what America can offer. Over time, the world decides it basically wants to deal with China. It wants to buy its weapons from China, its robots from China, and its pharmaceuticals from China. Even many of America’s former allies move much closer to the Chinese orbit. A country that has a lead in AI will be able to convert that into a lead in many other sectors over time.
Taiwan will surrender without China having to fire a shot, and over time U.S. forces will be pushed out of Asia. South Korea will fall under Chinese influence.
"China will push the world to become more autocratic and also more pro-Chinese," writes Cowen. "Chinese agentic swarms rule the internet, and any media outlet or politician that is too anti-China will face their cyber wrath."
Don't believe it? Who's going to stop them?
No one knows if an "AI doom scenario" is possible. "Is belief in it a good way to protect and extend life, liberty, and the pursuit of happiness?" says Cowen. "Will it help us much, or for long, if it is Chinese AI that turns on us and does us in?"
Perhaps before we scuttle AI data centers and force AI companies to handcuff themselves, we should ask what the real alternatives mean for the future of the United States. I don't see many opposing views that postulate an alternative scenario to "AI doom" where China is the major beneficiary of our hysteria.
