Businesses are rushing to adopt AI tools that they believe will significantly improve productivity. But one professor warned that technology could be quietly hollowing out the workforce instead.
Anastasia Berg, an assistant professor of philosophy at the University of California, Irvine, said new research and what she’s heard directly from colleagues across industries shows that workers who rely heavily on AI are losing core skills at an alarming rate.
“We have a tremendous amount of empirical data on the issue of skill attrition and skill atrophy,” Berg said on The Philosopher podcast this week. “We often talk about what it takes to acquire a skill,” but it’s also necessary to maintain the skill, she said.
Although Berg did not cite any specific studies, research from Oxford University Press and journals such as Springer and MDPI suggests that AI may increase the speed and motivation of learning, but often at the expense of depth, critical thinking, creativity, and long-term skill development.
AI may be harming the workers who need to learn the most
Berg said the workers most vulnerable to this deskilling effect are younger workers.
She said this is not just an issue in humanities subjects. Computer science professors say students and young developers are relying so heavily on AI tools that they are no longer able to learn how to write and debug code on their own.
“It’s another thing for senior programmers to use AI,” she said. But juniors can’t help but use it, so it’s no good.
Because they rely on AI from day one, Berg says, they never build the foundational knowledge needed to understand how the AI works, much less validate or modify it.
AI is becoming a crutch outside of work
Berg said reliance on AI extends far beyond the workplace. Adults are now turning to chatbots for everything from emotional support to everyday decision-making. She believes this change has undermined her ability to make independent judgments.
“The vast majority, if not even close, of AI use among adults is not work-related,” she said, citing “constant advice,” “a lot of weird sociability,” and “emotional task management.”
An analysis of 1.58 million ChatGPT conversations by researchers from OpenAI, Duke University, and Harvard University found that by June 2025, 73% of messages from adult users were not work-related, but the study did not analyze specific non-work uses.
Such dependence weakens the cognitive abilities people need to not only perform professional tasks but also function independently in everyday life, she said.
A looming competency crisis
Berg’s point is that AI doesn’t just automate tasks; it automates the very process by which people develop skills.
As workers become dependent on AI, they lose the friction that powers their reasoning, problem-solving, and decision-making abilities.
“We’re forcing them to sacrifice the most basic level of their abilities,” she said. “The threat to their highest level of competency is simply tremendous.”
If companies continue to implement AI into all workflows in the name of efficiency, they could create a generation of employees who may seem productive on paper, but who need digital hands-on work to perform at their best, he said.
In other words, AI may not be enhancing the workforce. It may be gradually disintegrating.
