A new study from MIT suggests that artificial intelligence could already replace nearly 12 percent of the U.S. labor market, affecting jobs that account for as much as $1.2 trillion in wages across finance, healthcare, and professional services.

The research, conducted with the Oak Ridge National Laboratory, used a labor simulation tool called the Iceberg Index to model how AI affects 151 million American workers. The index maps more than 32,000 skills across 923 occupations and measures where current AI systems can already perform those skills.

 Researchers found that the most visible job shifts in tech represent just a small fraction — 2.2 percent, or $211 billion — of the total wage exposure. The larger, submerged part of the “iceberg” includes $1.2 trillion in wages across functions like human resources, logistics, and office administration — areas often overlooked in automation forecasts.

Rather than predicting specific job losses, the researchers say the index provides a snapshot of AI’s current capabilities to help policymakers make informed decisions. Several state governments, including Tennessee, North Carolina, and Utah, have already partnered with the researchers to run simulations and build policy scenarios.

“Project Iceberg enables policymakers and business leaders to identify exposure hotspots, prioritize training and infrastructure investments, and test interventions before committing billions to implementation,” the report says.

One North Carolina state senator, DeAndrea Salvador, who worked on the project, highlighted the tool’s ability to provide localized data. “One of the things that you can go down to is county-specific data to essentially say, within a certain census block, here are the skills … and what could that mean in terms of the shifts in the state’s GDP in that area, but also in employment,” she said, according to CNBC.

The study also challenges the idea that AI’s impact will be limited to coastal tech hubs, showing that exposed occupations are spread across all 50 states.

Meanwhile, Geoffrey Hinton, one of the so-called “godfathers” of AI, continues to voice dire warnings about the technology. During a conversation with Senator Bernie Sanders at Georgetown University last week, Mr. Hinton argued that AI’s rapid deployment will be unlike past technological revolutions.

“The people who lose their jobs won’t have other jobs to go to,” Mr. Hinton said. “If AI gets as smart as people — or smarter — any job they might do can be done by AI.”

Mr. Hinton, who won a Turing Award for his pioneering work on the neural networks that underpin modern generative AI, has previously expressed regret over his life’s work. He believes the AI industry cannot be profitable without replacing human labor and has warned that we are less than 20 years away from artificial general intelligence, an AI with human or superhuman intelligence.

Strikingly, Mr. Hinton now claims that the latest models like the unreleased GPT-5 “know thousands of times more than us already.” However, some experts disagree, pointing out that while large language models are trained on vast amounts of data, they don’t truly “know” or understand the information. Furthermore, attempts to replace workers with AI agents have often failed, particularly in customer service roles.

Despite these setbacks, Mr. Hinton remains concerned. He argued that billionaires like Elon Musk and Mark Zuckerberg haven’t considered the economic consequences of mass unemployment. “If the workers don’t get paid, there’s nobody to buy their products,” he said.