Workers who are using AI to do their job are rated as less competent than those who are not, even if the AI users are transparent about their methods, and even if the quality of the work done with AI is identical to the quality of that done without, new research suggests.
Four academics, from King’s Business School in London, Peking University, the Hong Kong Polytechnic University, and University of Hong Kong conducted an experiment in which they got 1,026 engineers to evaluate a Python code snippet that was written by another engineer—either with or without help from AI.
In every case the code itself was identical; only the described method of creation differed from sample to sample. Yet when reviewing engineers believed that an engineer had used AI to write the code, they rated that engineer’s competence an average of 9% lower.
Perhaps even more striking: this competence penalty was more than twice as harsh for female engineers than it was for male engineers. Indeed, the women in the study faced a 13% reduction compared to the 6% reduction suffered by their male counterparts. “When reviewers thought a woman had used AI to write code,” the researchers explained in the Harvard Business Review, “they questioned her fundamental abilities far more than when reviewing the same AI-assisted code from a man.”
The penalties also varied significantly depending on the evaluator. Engineers who had not adopted AI themselves were the harshest critics—especially male non-adopters evaluating female engineers. In those cases, they penalized women 26% more than men for the exact same AI usage.
Reflecting on their findings, the researchers noted that they reveal a “hidden tax” on AI adoption. “What looks like simple reluctance to use new tools actually reflects rational self-preservation. The true cost extends far beyond lost productivity—although productivity losses alone can also be substantial,” they wrote.
It also casts a fresh light on the gender gaps that permeate the evolving workforce—from trust to perceived competency—and underscores the risks of new technologies exacerbating existing inequalities, such as pay gaps.