AI adoption rates show no sign of slowing. Organisations are rushing to adopt state-of-the-art AI coding technologies for their employees, with 78% of businesses in 2025 using AI in at least one function. However, amid the pace of transformation, a hidden stigma is preventing employees from embracing it fully in their day-to-day roles.
Recent research speaks to this problem. A recent Pew Research Centre survey revealed that 91% of American workers are allowed to use AI – but only 16% actually did. In some cases, this might be due to skills gaps or a lack of awareness of where AI could help them. However, we’re increasingly seeing workers hesitating to use the tool because they fear how they will be perceived if ‘caught’ using it.
Harvard Business Review’s latest findings on the ‘hidden penalty of using AI at work’ unveiled similar insights. The publication surveyed engineers at a leading tech company, only to realise that less than half of the company’s engineers had been using the AI tools it offered. To understand why this might be, participants evaluated code written by another engineer, either with or without AI assistance. Engineers who are believed to use AI are deemed, on average, 9% less competent by their peers.
This unveils an unfortunate but unsurprising truth: competence penalty is still rampant in the tech world. Perhaps most alarming, the problem is twice as severe for women. Those deemed to have used AI faced a 13% reduction, compared to 6% for male engineers.
Women are disproportionately affected by the competence penalty, often judged as less capable simply for leveraging the tool. Now this bias is showing up in how they, and other employees, might use AI in the workplace. Understandably, women might therefore hesitate to use AI for coding – if they do use it, it might be used as ‘proof’ against them that they can’t do the work otherwise.
This is echoed by the findings: engineers who hadn’t adopted AI were the harshest critics of those who had used it. Specifically, male non-adopters evaluated code by women who’d used AI 26% more harshly than male engineers who had.
The competence penalty and negative perceptions around AI will do more than just knock women’s confidence – it’ll create larger barriers to adoption of new technologies. Some may put off using the tools as a self-preservation instinct, reinforcing inequality between employees and impeding adoption within the business. They’ll also be impacted negatively, as their AI investment won’t yield the return expected.
It is in an organisation’s best interest to address prejudice against AI adoption. It can play a key role in normalising it as a tool that enhances productivity, not one that replaces competence. This includes creating learning environments where people can experiment with AI without judgment, and ensuring that employees are not penalised for using sanctioned tools.
Unfortunately, the longstanding bias against women in STEM is nothing new and predates the advent of AI. GitHub’s 2016 study – nearly ten years ago – revealed a similar sentiment. Three million code submissions were presented without disclosing the engineer’s gender. Anonymised female-written code saw a 78.6% approval rate, but shot down to 62.5% when gender was announced.
Business leaders have a responsibility to address this prejudice and help women embrace new technology without fear of judgment. Organisations need to find creative ways to get more women involved and to support them. On a day-to-day level, this can look like organising hackathons and workshops to get teams excited. These efforts, however, should be holistic: business leaders need to ensure that the culture around technologies is inclusive and that female employees can give feedback on their experience.
Simone Mink is the product operations lead at Mendix, a Siemens business.
Read more: Mergers and acquisitions make your tech stack look mysterious. That’s not a good thing.
