Many workers fear artificial intelligence and see it as competition in the labor market. According to the latest MIT research, these concerns are not unfounded – they reflect a real problem already affecting tens of thousands of jobs.
According to the study, which analyzed data from more than 800 occupations, AI systems are now advanced enough to perform a share of tasks classified as “routine cognitive” – such as preparing reports, processing data, checking documentation quality and overseeing simple workflows. Yet despite this potential, the pace of adoption remains much slower than media narratives about “mass automation” suggest.
MIT estimates that in current market conditions, replacing human labor with AI is often not financially viable because the implementation, integration and oversight costs exceed short-term savings. As a result, although AI could theoretically replace 11.7% of jobs, real-world automation is so far concentrated in narrow domains: customer support, simple accounting processes and selected office functions.
The researchers also note that these costs decrease each year – meaning the pace of automation could accelerate significantly, especially in occupations with high repetition and poorly defined career progression. Some industries, such as insurance, logistics and public administration, may be among the first to experience rapid AI deployment.
Importantly, the authors stress that automation does not always mean immediate layoffs – in many organizations, AI is used as an assistive tool that takes over monotonous tasks, while workers shift to more strategic or creative roles. MIT also predicts that new positions related to supervising, fine-tuning and maintaining AI systems will offset at least part of the job losses.
The MIT study fits into a broader debate about the future of work in the age of AI. Experts note that the coming years will depend not only on technological capability, but also on regulatory decisions, energy costs, market pressures and corporate productivity strategies. The findings suggest that the true employment disruption will occur only when automation becomes clearly cheaper than human labor which, according to the researchers, may happen in the second half of this decade.