AI Isn't 'Revolutionary Change,' and Its Benefits Are 'Exaggerated,' Says MIT Economist Less than 5% of human jobs will be affected by AI, he claims.

An award-winning economist is calling out what he sees as an overinvestment in AI — and he's issuing a warning that overhyping machines and underrating humans could be a big mistake.

In a Tuesday interview with NPR, MIT economist Daron Acemoglu explained that he didn't think AI would revolutionize the economy in the next 10 years.

"I mean, unless you count a lot of companies overinvesting in generative AI and then regretting it," he said.

Acemoglu, who won the Global Economy Prize in 2019, believes that AI's effects are "exaggerated" and that the technology will not be able to perform many tasks outside of an office. Even inside an office, he says, AI can't fully replace humans because it still makes mistakes and relies heavily on the data it has been trained on, which may have been copyrighted.

Related: How Much Does It Cost to Develop and Train AI? Here's the Current Price, According to the CEO of an $18 Billion AI Startup.

Acemoglu wrote a paper in April that measured AI's long-term impact on the economy. He found that less than 5% of human jobs will be affected by AI and that AI will only have a "modest" impact on GDP over the next decade.

The question remains if we will ever need ChatGPT to "write Shakespearean sonnets" as it does now, when "what we want is reliable information useful for educators, healthcare professionals, electricians, plumbers, and other craft workers," the paper reads.

AI's accuracy has been repeatedly questioned, with Google's AI overviews getting major things wrong within a month of release and NYC's $600,000 AI chatbot spotted giving business owners inaccurate advice.

^{Daron Acemoglu. Credit: Frank Molter/picture alliance via Getty Images}

Acemoglu told NPR that AI companies have used his books and academic papers to train AI, even though he didn't give them permission. He tried summarizing his work with AI out of curiosity and said it was "not horrible," but that a human being hosting the podcast could do it better.

"A lot of people in the industry don't recognize how versatile, talented, multifaceted human skills - capabilities are," Acemoglu said. "And once you do that, you tend to overrate machines ahead of humans and underrate the humans."

Related: Microsoft AI CEO Says Almost All Content on the Internet Is Fair Game for AI Training