11.7% of jobs could be replaced by artificial intelligence

Redazione RHC:28 November 2025 16:08

Artificial intelligence is currently capable of performing work equivalent to 11.7% of US employment . And not just in theory: in terms of monetary value, this value is already comparable to the salaries of human workers.

An MIT study , developed as part of Project Iceberg , shows that, at the current level of technological development, it is possible to automate tasks worth approximately 1.2 trillion dollars, a market share in which artificial intelligence can already perform the same functions as humans, while costing less.

Unlike previous estimates, which were based on assumptions about the “susceptibility of professions to automation,” the researchers focused on another aspect: where exactly AI performs tasks as well as humans without incurring additional costs for employers. This is precisely what determines the 11.7% figure: not as an illusion for the future, but as a very realistic amount of work that can be outsourced to algorithms already now.

Project Iceberg was developed in collaboration with Oak Ridge National Laboratory, home of Frontier , one of the world’s most powerful supercomputers. It was used to create a detailed simulation of the US labor market: 151 million virtual workers with diverse occupations, skills, and locations . The system takes into account 3,200 skills across 923 job titles, covering 3,000 counties across the country. This digital model was compared with the capabilities of modern artificial intelligence systems, from language models to specialized software.

It’s important to understand that the study doesn’t predict that 11.7% of jobs will disappear in the next year. This means that artificial intelligence can already replace humans in these tasks, both technically and economically. However, the gap between potential and actual practice remains wide.

AI is currently widely used in IT, primarily for programming. These activities account for approximately 2.2% of salaries, or approximately $211 billion annually. However, researchers estimate that much greater automation potential lies in other sectors, particularly those previously considered beyond the capabilities of machines. These include routine tasks in finance, healthcare, logistics, human resources, accounting, and law— professions long considered relatively immune to automation.

The main risk, it seems, lies not even in production lines, but in office routines, where repetitive tasks related to document management, calculations, processing requests, and other typical administrative functions are performed daily. Artificial intelligence is already addressing this problem, if not everywhere, then at least in a significant number of cases.

But this doesn’t mean that mass migration of people is a matter for the coming years. Previous research from MIT (particularly the CSAIL laboratory) shows that, even if technology can replace humans, it doesn’t always make sense. Model training, process redesign, and organizational costs are often prohibitive. Another study, this time from MIT Sloan, found that the implementation of AI did not lead to net job losses between 2010 and 2023. On the contrary, in many cases, companies grew, and along with revenue, so did the number of employees.

The index itself, calculated as part of Project Iceberg, is not necessary for layoff forecasting. It is intended as a planning tool: it allows for modeling different scenarios and understanding in advance where workforce training needs to be strengthened, infrastructure overhauled, or new support measures developed. Some US states, such as Tennessee, North Carolina, and Utah, are already using this platform. These states are using the calculations to formulate regional strategies for adapting to AI.