If an AI tool makes your job easier, does it quietly make you worse at catching its mistakes?

This explores what happens to workers' skills, their ability to oversee AI, and their leverage at work when they lean heavily on AI tools that are supposed to help them, not replace them. The usual comfort is that augmentation is the safe path: if AI assists rather than automates, humans stay in charge. The corpus pushes back on that. A study that mapped more than 8,000 workplace AI risk scenarios found that augmentation offers no built-in protection. Over time, heavy reliance can wear down the skills workers need to judge whether the AI is right, which undercuts the main reason augmentation was called safe in the first place Does AI augmentation protect workers from skill erosion?.

The erosion is not uniform, though. Interviews with Dutch knowledge workers turned up four outcomes, not a simple upskilling-or-deskilling split: some skills develop, some hold steady, some fade, and some get revalued as their importance changes. Which one happens depends on how a person uses the tool and which parts of the job it touches How does generative AI actually change worker skills?. Self-reports also make the picture murkier. Anthropic found that the people who delegate the most to Claude are the most optimistic about their careers and believe their skills are gaining value Does delegating work to AI actually damage worker skills?. That finding is correlational and comes from Anthropic's own users, and confidence is a different thing from competence. Overreliance may be hardest to notice in exactly the people most exposed to it.

A less obvious mechanism is that overreliance hides the work itself. An analysis of 1,250 interviews found that workers carefully keep their own voice and attribution in AI-assisted output, but let signs of effort, attention, and uncertainty disappear into the finished product Which workplace cues survive AI mediation and which disappear?. Once a polished deliverable counts as proof that work happened, nobody looks closely at whether anyone actually thought it through, including the worker. That makes skill erosion hard to detect from the outside.

The cost becomes visible when the AI fails and a human has to take over. In a randomized experiment on Alibaba's Taobao, workers who took over from failing customer-service AI restored quality for technical problems but not for frustrated customers, and late handoffs rarely recovered at all When human workers take over from AI, does it actually fix the problem?. Human backup only works if the human is ready and steps in early. The same setting also shows that reliance can go well: workers supervising the AI redirected their freed-up attention and improved on the chats they still handled fully themselves Does supervising AI free up worker attention for other tasks?. Whether reliance turns into neglect or into reallocated attention seems to depend on whether there is meaningful work left to put that attention toward.

The stakes go beyond individual skills. If workers can no longer do or check the work, the line between augmentation and automation blurs, and Brynjolfsson warns that automation moves bargaining power to whoever controls the technology, which can lock workers into a position with little leverage Does automating labor concentrate power more than augmenting it?. At the scale of whole societies, the gradual-disempowerment argument holds that institutions stay aligned with human interests partly because they depend on people who care about the outcomes. Remove that dependence one task at a time and the drift away from human preferences could become irreversible Does incremental AI replacement erode human influence over society?. The surprise is that overreliance is less about one worker getting rusty and more about quietly removing the human check that both the workplace and the wider system were counting on.