Artificial intelligence is starting to blur the line between what workers can produce and what they can independently do, creating a new challenge for companies trying to evaluate skills, promotions and performance.
Euronews reported that a Use.AI survey of 9,684 working adults across the U.S., UK, Canada, EU and Latin America found 52% believed AI had made them appear more experienced than they actually are.
The same research found that 64% had used AI to complete work they could not otherwise have done independently, while 43% said the technology helped them take on responsibilities they did not yet feel qualified to handle.
Verity similarly reported that more than half of surveyed workers said AI made them appear more experienced, highlighting a widening gap between polished output and underlying experience.
AI-assisted work is increasingly invisible
The challenge for employers is that AI use is not always disclosed.
According to Euronews, 39% of respondents had submitted AI-assisted work without mentioning the technology, while 30% said they had accepted praise for work substantially produced with AI.
Another 35% said they would struggle to perform parts of their current job without AI, while one-quarter worried their employer believed they were more capable than they really were.
The consequences can extend into career progression, wherein 19% said AI-assisted work had contributed to a promotion.
Ihor Herasymov, co-founder and CEO of Use.AI, argued that companies should not require disclosure for every AI interaction. Instead, he shared that disclosure should focus on cases where AI “materially shaped the substance of the work”.
Output may no longer measure capability
The findings create a difficult product-management and hiring problem: finished work increasingly reveals less about how much of the underlying task an employee actually understands.
Herasymov said that employers may need to evaluate whether workers can explain their reasoning, detect flaws in AI-generated answers and respond when the technology is wrong, rather than relying solely on polished deliverables.
That distinction matches broader research into how workers want AI deployed.
Stanford HAI reported that a study of 1,500 workers across 104 occupations found 45.2% preferred an equal partnership between humans and AI, while another 35.6% wanted human oversight at critical points.
The Stanford research also found that 69.4% welcomed automation that freed time for higher-value work, but 45% expressed doubts about AI accuracy and reliability.
Companies may need to redefine what skill means
For engineering and product leaders, the issue is bigger than whether employees should disclose using ChatGPT or another assistant.
AI can legitimately extend what a worker can accomplish, much as software tools have always augmented professional ability. The harder question is whether companies are evaluating tool-augmented performance or transferable human capability.
That matters when promoting someone into a role where they must diagnose failures, challenge incorrect outputs or make decisions when AI cannot provide a reliable answer.
As AI becomes embedded into everyday workflows, companies may therefore need to measure less of what workers can produce with assistance—and more of whether they understand, verify and defend the work they ultimately put their name on.
