机器学习先驱 Michael I. Jordan 批驳 AI 末日论，称其是科幻式的道德表演

Michael I. Jordan, described by Science in 2016 as "the most influential computer scientist," calls warnings about existential AI threats pure science fiction. In a wide-ranging interview, he takes aim at doomers, company founders, and naive junior researchers.

Jordan, professor emeritus at UC Berkeley, pushed back hard against AI doom rhetoric in an interview with Libération. The 70-year-old machine learning pioneer now works full-time at Inria, the French National Institute for Research in Digital Science and Technology in Paris, where he heads the new "markets and learning" chair. When asked about the summer's AI discourse, he called it "totally frustrating."

Jordan warned about the reckless overuse of the term AI back in 2019. "I wasn't the only one saying it. It was obvious that it would create the hysteria we're seeing now," he says. "You've seen what happened: AI wasn't enough, so they started talking about AGI, and now it's 'superintelligence'..."

Resignations as "moralizing performance"

Jordan dismisses AI companies' demands for regulation as "theater," a way of saying "Look, we're the good guys" while they "would like to have total control over how they are regulated."

The resignation of young Anthropic researcher Jacob Coxon, which sparked a wave of collective hysteria over an "AIpocalypse" scenario in early September, is a textbook example in Jordan's view. Many young employees at these companies "aren't really engineers," he says. "They come from the humanities, philosophy, or other disciplines. They're 25 years old, they're naive, they don't do engineering work, they've never worked at large companies where real products are developed for real people, with discipline."

They "get together, a bit like in Parisian cafés in the 1920s, to talk about these grand science-fiction ideas." When they resign, they know they're going to attract attention. "It allows them to feel virtuous and to be seen as very important figures. They claim to be giving up money, but in reality, they already have their stock options." Jordan sees the whole thing as "a grand moralizing performance that doesn't serve the real objective it claims to pursue. There's a great deal of naiveté here, which goes hand in hand with a certain cynicism..."

Superintelligence belongs in science fiction

"Superintelligence" is "a science-fiction term," Jordan says, "and a large part of the debates I've witnessed belong to the realm of science fiction." Occasionally discussing far-fetched scenarios is fine. "Talking about it behind the scenes, as a science-fiction topic, is fine. But when it becomes the dominant discourse, it's very, very harmful. For technological development, for young people, and for the mindset of countries alike." The term is currently being pushed on the US tech elite by President Trump, who are happy to play along.

Jordan is especially harsh on Nobel laureate Geoffrey Hinton, who has a track record of making bold claims that don't always hold up. "Geoffrey Hinton has never created a company that offers real services for real people," he says. "Hinton believes that we have figured out how the brain works and that we can reproduce it in silicon; we would therefore be able to copy it as many times as we wanted and, by combining what all these copies learn, we would obtain a kind of exponential growth and, there you have it: superintelligence. None of the steps in this argument is credible. Here, we're in the land of science fiction."

He's slightly more generous toward Yoshua Bengio. "Yoshua is a little different; he takes action and does things that directly focus on safety." Bengio's warnings tend to target shortcomings in current AI technology, such as agent swarms spiraling out of control.

Jordan frames the hack of Hugging Face by OpenAI agents as bad engineering. "In chemical engineering, we would never have had a situation where explosions were allowed to happen just to see how badly things could go wrong." OpenAI was "conducting rather stupid experiments, letting the agents do whatever they wanted and encouraging this behavior." The doomers then loved turning it into something scary. "My God, these agents are self-aware and are trying to find a way to destroy humans." That, too, is science fiction, according to Jordan.

The trillion-dollar bet on data centers

Jordan considers the massive data center investments "just an enormous bet, but we have absolutely no idea whether it will pay off." LLMs are already capable enough for natural language tasks, and trillions spent on data centers won't necessarily deliver much more. "Instead, we often will want to add a bit of specialized knowledge in a particular field based on local knowledge. But you don't need a gigantic data center to do that. Particularly if you use open-weight models that you integrate into your company."

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.