Study Finds AI Assists Allow Workers to Appear More Skilled Than They Are

A new study published Tuesday by Use.AI indicates that a significant portion of office workers believe artificial intelligence allows them to present a false image of their professional capabilities. The research, which surveyed 9,684 working adults across the US, UK, Canada, the EU, and Latin America, found that 52% of respondents felt AI made them appear more experienced than they actually are.

The data highlights a growing gap between delivered results and underlying skill. Approximately 64% of workers reported using AI to complete tasks they could not have handled independently, while 43% admitted it helped them take on responsibilities for which they did not yet feel qualified. Furthermore, 35% conceded they would struggle to perform parts of their current roles without the technology.

Transparency appears to be lacking in many workplaces. The study found that 39% of employees had submitted AI-assisted work without disclosing its use, and 30% had accepted praise for outputs substantially produced by the technology. Consequently, 25% of workers worried their employers overestimate their actual capabilities. These dynamics have tangible career consequences, with 19% of respondents stating that AI-assisted work contributed to a promotion.

Ihor Herasymov, co-founder and CEO of Use.AI, argued against mandatory disclosure of every AI interaction, noting that such a policy would become impractical as the tools become embedded in everyday software. Instead, he proposed a threshold based on material assistance: employees should disclose when AI generated a significant part of an analysis, recommendation, presentation, or code.

“The intent should not be to police staff,” Herasymov told Euronews. “The purpose should not be surveillance. It should give managers enough context to judge both the work and the human contribution to it.” He emphasized that while AI assistance should be noted, the person submitting the work remains responsible for understanding and defending it.

The report suggests that traditional metrics of performance are becoming less reliable. Finished output no longer reveals as much about an individual’s capabilities as it once did. Herasymov advised that employers need to assess whether employees can explain their reasoning, detect flaws in AI-generated answers, and make sound decisions when information is incomplete or the technology fails.

“Problem framing is particularly important,” Herasymov stated. “Can someone define the right question, challenge an assumption and explain why one course of action is better than another? That is much harder to infer from a polished final output alone.”

Addressing concerns that heavy reliance on AI might lead to dependency rather than augmentation, Herasymov acknowledged the validity of the issue. “The concern begins when someone can produce an answer with AI but cannot reliably recognise when that answer is wrong,” he said. “At that point, the relationship starts to look less like augmentation and more like dependency.”

The urgency of this assessment is compounded by the rapid advancement of AI autonomy. Citing data from METR, the article notes that the time required for AI autonomy to double has compressed from an eight-month trend to just 4.7 months. Between early 2025 and early 2026, AI’s autonomous capabilities grew by 1,400% year-on-year. Additionally, downloads of AI tools surged from 15,000 to 11.8 million—a 780-fold increase—while publicly available Model Context Protocol (MCP) tools grew 35-fold to approximately 177,000.

MCP standards allow AI assistants to plug directly into applications and execute tasks, a development known as “agentic AI.” As these technologies enable more work to proceed without direct human direction at every step, the final products will reveal progressively less about the individual who signed their name to them.