MIT puts a number on how many US workers AI is already capable of replacing

A new study from the Massachusetts Institute of Technology puts a number on how many US jobs could already be done by artificial intelligence (AI). The study, which CNBC has seen, finds that AI can already replace work completed by 11.7% of the US workforce, or about $1.2 trillion in wages across major sectors like finance, healthcare, and professional services. The research used a new tool, the Iceberg Index, developed by MIT and Oak Ridge National Laboratory (ORNL). The index simulates how 151 million US workers interact across the country and how AI and government policies might impact them.The Iceberg Index offers a detailed look at how AI may change the job market-not just in technology centres but in every state, down to the zip code level. For government leaders planning major investments in worker retraining and education, the index provides a map of where job disruption is likely to occur.Talking about the study using the Iceberg Index, Prasanna Balaprakash, ORNL director and co-leader of the research, said: “Basically, we are creating a digital twin for the US labour market.”ORNL is a Department of Energy research centre in eastern Tennessee, home to the Frontier supercomputer, which powers many large-scale modelling efforts.The index runs population-level experiments, showing how AI changes tasks, skills and labour flows before those changes appear in the real economy, Balaprakash said. The index treats 151 million workers as individual agents, each tagged with skills, tasks, occupation and location. It maps more than 32,000 skills across 923 occupations in 3,000 counties, then measures where current AI systems can already perform those skills.Researchers found that the visible part of layoffs and role shifts in tech, computing, and information technology accounts for just 2.2% of total wage exposure, or about $211 billion. Below the surface lies the total exposure: $1.2 trillion in wages, including routine functions in human resources, logistics, finance, and office administration.However, it's important to note that the index is not a prediction tool about when or where jobs will be lost. Instead, it gives policymakers a structured way to explore different scenarios before committing real money and enacting legislation.The researchers partnered with Tennessee, North Carolina and Utah, which helped validate the model using their own labour data. Tennessee cited the Iceberg Index in its official AI Workforce Action Plan released this month. Utah is also preparing a similar report.North Carolina state Sen. DeAndrea Salvador said the ability to drill down to local detail is helpful."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 that is currently happening now and then matching those skills with what are the likelihood of them being automated or augmented, and what could that mean in terms of the shifts in the state's GDP in that area, but also in employment," Salvador said.The Iceberg Index even shows the exposed occupations spread across all 50 states, including inland and rural regions."Project Iceberg enables policymakers and business leaders to identify exposure hotspots, prioritise training and infrastructure investments, and test interventions before committing billions to implementation," the report noted.