DeepMind co-founder Mustafa Suleyman is a heavyweight in the AI space. The Oxford dropout worked as a negotiator for the United Nations and the Dutch government early in his career, but then pivoted to AI and founded DeepMind in 2010 alongside Demis Hassabis and Shane Legg.

The machine learning lab grew like a weed under Suleyman, with the backing of Peter Thiel’s Founders’ Fund, before selling to Google parent company Alphabet for £400 million in 2014. Suleyman then took on several roles at DeepMind before stepping down five years later.

Now, the veteran AI founder is working on a new company called Inflection AI, which offers personalized AI assistants. And while Suleyman remains an avid supporter of AI, he expressed concerns about the industry’s possible negative effects—in particular on workers.

“In the long term…we have to think very hard about how we integrate these tools, because left completely to the market and to their own devices, these are fundamentally labor replacing tools,” Suleyman told CNBC on January 17 at the World Economic Forum’s annual gathering in Davos, Switzerland.

AI tools do two main things fundamentally differently, the DeepMind co-founder said. First, they make existing operations more efficient, which can lead to huge savings for businesses, but often by replacing the humans who did those jobs. Second, they allow for entirely new operations and processes to be created—a process that can lead to job creation. These two forces will both hit the labor market by storm in coming years, leaving a serious, but unpredictable impact.

While Suleyman expects AI to “augment us and make us smarter and more productive for the next couple decades,” over the long term, its impact is still “an open question.”

Experts have been debating whether AI will replace human workers for over a decade. Some researchers argue that AI will lead to a wave of unemployment and economic disruption as it takes jobs worldwide, but others believe that the technology will create new job opportunities and spur economic growth by boosting worker productivity.

There’s been a steady stream of academic papers on the topic. A 2013 study by Carl Benedikt Frey and Michael Osborne, for example, estimated that 47% of US jobs are at risk of being automated amid the AI boom by the mid-2030s. And a July McKinsey study found that nearly 12 million Americans will need to switch jobs by 2030 as AI takes over their roles.

On the other hand, some researchers have found that AI could boost economic growth and offer new opportunities for workers. A 2022 United Nations’ International Labor Organization (ILO) study found that most AI systems will complement workers, rather than replacing them.

Still, Suleyman isn’t the only big name in the AI industry to warn about the scary implications of AI for the labor market.

In a Jan. 10 Wired article, MIT professor Daron Acemoglu predicted that AI would disappoint everyone in 2024, proving itself merely a form of “so-so automation” that will take jobs from workers but fail to deliver the expected monumental improvements to productivity.

Researchers have yet to solve the problem of hallucinations—where generative AI systems exaggerate or fabricate facts—and that could lead to a whole host of issues in coming years, the noted economist argued, adding that there’s “no quick fix” to the problem.

“Generative AI is an impressive technology, and it provides tremendous opportunities for improving productivity in a number of tasks. But because the hype has gone so far ahead of reality, the setbacks of the technology in 2024 will be more memorable,” Acemoglu wrote.

For Suleyman, unlike Acemoglu, it’s not that hype surrounding AI isn’t real, it’s definitely a “truly transformational technology.”

“Everything that is of value in our world has been created by our intelligence, our ability to reason over information and make predictions. These tools do exactly that, so it’s going to be very fundamental,” he explained on January 17.

Suleyman instead fears that AI will be so good at replicating humans that it will eventually displace workers, and without regulation, that could lead to serious economic consequences.

That being said, like Acemoglu, Suleyman argued that AI’s proponents might be getting ahead of themselves with their optimistic near-term outlooks for rising productivity. The true impact of AI, from its ability to birth revolutionary technologies to its potential to stoke epic job losses, likely won’t hit for years.

“AI is truly one of the most incredible technologies of our lifetimes, but at the same time, it feels like expectations about its delivery are higher than they’ve ever been and maybe we have hit a kind of peak hype for this moment,” he explained.