Silicon Valley has a plan to save humanity: turn on the nuclear reactors

() – Artificial intelligence did not become the utopia that technology evangelists anticipated. Until now, artificial intelligence has proven to be more capable of generating stock market enthusiasm that, for example, great tangible things for humanity. Unless you count Shrimp Jesus.

But that’s going to change, AI advocates tell us. Because the only thing standing in the way of an AI-powered idyll is lots and lots of computing power to train and operate these nascent AI models. And don’t worry, friends of the public who never asked for any of this: that power will not come from the fossil fuels. Imagine the public relations headaches.

No, the technology that is going to save humanity will be powered by the technology that almost destroyed it.

This is the deal: To do AI on the scale that the Microsofts and Googles of the world envision, you need a lot of computing power. When you ask Chat-GPT a question, that question and its answer are consuming electricity on a supercomputer full of Nvidia chips in some remote, air-conditioned data center.

The power consumption of data centers, AI and cryptocurrency mining (its own environmental headache) could double by 2026according to the International Energy Agency.

In the United States alone, energy demand is expected to grow 13% to 15% annually through 2030, which could make electricity a much scarcer resource, according to analysts at JPMorgan.

The technology industry’s solution, for now, is nuclear energy, more stable than wind or solar and practically free of carbon emissions.

The irony of all this, of course, is that even AI advocates invoked the history of nuclear proliferation to try to convey the need to put up barriers to artificial intelligence (as long as regulations don’t slow it down or reduce its benefits). by no means).

And while predictions about the future of artificial intelligence are often dismissed as alarmist, those who worry about nuclear power cannot be so easily dismissed. History is, tragically, on their side.

Anna Erickson, professor of nuclear sciences at Georgia Tech, told me that nuclear energy is better understood today than in 1979, when reactor 2 at Three Mile Island suffered a partial core meltdown.

“Nothing in life is foolproof,” he says, “but we now have a much better understanding of how nuclear reactors work,” thanks in part to the wave of safety regulations triggered by the Three Mile Island incident.

Conclusion: There is no future for AI without a major increase in our energy supply, making the expansion of nuclear energy virtually inevitable. But many of the recently announced projects will take years to come online, meaning big tech companies’ data centers will have to continue burning fossil fuels as long as demand continues to rise.

Are we okay with destroying the planet if all we get are applications capable of summarizing our emails? Or search engines that are a little more human but less reliable? Is the future really just variations of crustacean-based deities in a smoothie of AI slop?

There is a lot at stake, including our jobs, the environment and our entire sense of the world, according to the AI ​​developers themselves. And yet, it remains unclear what people will get out of all this.