AI use at work raises risks for skills, trust and career growth

Artificial intelligence is now part of daily work for 43% of U.S. employees, according to a March study from the Federal Reserve Bank of St. Louis. Academics who study work, business and psychology say the career impact depends less on whether employees use the tools than on which tasks they hand over.

Sandra Matz, a professor at Columbia Business School, said AI can make workers more productive when it speeds up routine work or serves as a partner for testing ideas. The risk, she said, comes when employees ask the technology to do the thinking that builds expertise.

Matz said workers who use AI to replace critical judgment or creative reasoning may become complacent and fall behind. Because many AI systems tend to produce safe, conventional responses, she said, heavy reliance can push employees toward similar-sounding output and weaker independent thinking.

Productivity tool or substitute for judgment

Fred Oswald, an industrial-organizational psychologist and former member of the National Artificial Intelligence Advisory Committee, said employees and companies are still working through a “Wild West of AI use.” He said that creates a need to understand which applications reduce risk and which create it.

Americus Reed II, a marketing professor at the Wharton School of Business, said some of his colleagues refer to excessive reliance on AI for reasoning as “cognitive surrender.” Reed said AI is better suited to summarizing material, analyzing data and arranging ideas, rather than deciding what those ideas should mean.

“It should remove friction, not replace judgment,” Reed said, adding that the objective should be to spend less time handling information and more time creating meaning from it.

Dorothy Leidner, a business and ethics professor at the University of Virginia’s McIntire School of Commerce, said junior employees face a particular risk. Workers who have not yet built deep expertise may find it harder to use AI as a tool for extending thought, rather than as a generator of the thought itself, she said.

AI-written messages may weaken trust

Matz also warned against overusing AI for everyday communication with managers, employees and colleagues. She said AI can be useful for meeting summaries or progress reports, but informal exchanges at work carry signals that polished machine-written messages may remove.

If colleagues believe a routine message was produced by AI, Matz said, they may question whether the sender cared enough to write personally. She said imperfections, tension and spontaneity can help build stronger workplace relationships, and preserving those human cues will be a challenge for organizations adopting AI.

Leidner raised a related concern about relying heavily on “digital twins,” or virtual versions of a person, in communication with people they know. If AI writes messages and AI answers them, she said, the exchange no longer represents actual communication between people.

Identity and expertise under pressure

Eva Selenko, a work psychology professor at Loughborough University in England, said she is most concerned when employees use AI for tasks central to how they understand themselves professionally. Seeing a tool outperform a person on work that forms part of their expertise can become a status threat, she said.

Reed said workers may question their own value when AI performs tasks they long considered central to their role. He said the answer is not to avoid the technology, but to anchor professional identity in areas AI does not replicate as well, including judgment, creativity, curiosity and relationships.

This story draws on original reporting from CNBC.