Griffin, the hedge fund billionaire who manages one of the world’s most powerful trading firms, had long been one of the most prominent skeptics of AI in finance. As recently as January 22 at the World Economic Forum in Davos, he told a panel that while AI might look impressive on the surface, the moment you dug beneath it, “it’s all garbage.”

The reversal came fast—and, by Griffin’s own account, hit hard.

What changed his mind

Earlier this month, in a conversation at Stanford Business School, Griffin described a jarring personal reckoning. “I’ve got to tell you, I went home one Friday, actually fairly depressed,” he said. “You could just see how this was going to have such a dramatic impact on society.”

The catalyst was watching what AI was actually doing inside Citadel. Griffin said the technology had become “profoundly more powerful” than it was just months ago, enabling the firm to “unleash” a broader range of use cases it hadn’t previously been able to pursue.

The Stanford remarks came shortly after Griffin spoke at Milken Institute’s Global Conference, where he described asking CEOs to share how they were using AI to transform their businesses—and said he received “six or seven extraordinary stories” in response.

The finance jobs reckoning

For Griffin, the most striking proof isn’t in coding or content—it’s in high-end financial research. Work that Citadel would previously have assigned to teams with master’s degrees and PhDs in finance, work that took weeks or months, is now being completed by AI agents in hours or days.

“To be blunt, work that we would usually do with people with master’s and PhDs in finance over the course of weeks or months is being done by AI agents over the course of hours or days,” Griffin said at Stanford.

He drew a distinction between the more modest productivity gains AI has delivered in software engineering—where he pegged improvements at 15% to 25%—and the far more disruptive shift happening in knowledge work and research. “When you’re seeing really high-level research being done by AI engines, it’s quite eye-opening,” he said.

Griffin’s reversal is particularly notable given how recently he was pushing back on Wall Street’s AI boosterism. At Davos in January, he warned that the prediction that 50% of entry-level jobs would disappear within five years was “hype pushed to justify data center spending,” LinkedIn commentary on his remarks. He also noted that U.S. data center spending could top $500 billion this year, suggesting the investment narrative was outpacing real-world results. This is when he made the remarks that AI seemed to be “garbage” under the surface. What a change a few months make.

The broader warning

Griffin is now pointing to something further up the skills ladder—the automation of work once thought safe behind the wall of advanced degrees and specialized expertise. His message to workers is both urgent and pointed: adaptability is now the only durable edge.

“The success in your career will be defined as to whether or not you will be a lifelong learner or not,” Griffin said at Stanford, “and AI will just make this all the more important.”

It’s an open question whether the audience of young adults will be receptive to Griffin’s remarks. Nearly simultaneous with Griffin’s change of heart, commencement speakers have been getting booed by Gen Z audiences, seemingly whenever they mention the benefits of AI.

For this story, Fortune journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing.