AI is often promoted as a tool that makes workers more efficient, creative, and capable—helping them write faster, analyze data better, and complete tasks with less effort. But innovation theorist John Nosta, founder of the NostaLab think tank, warns that this promise comes with a hidden cost: over time, reliance on AI can erode human skills and judgment, leading to a decline in performance when the technology is no longer available. Nosta describes this phenomenon as the "AI rebound effect." He explains that while AI can significantly enhance performance in the short term, it may ultimately leave users less competent than they were before, even when they’re not using the tool. He illustrates this with a medical analogy: a doctor using AI to assist during a colonoscopy may detect more polyps and perform better with the technology. But the next day, without AI, the doctor’s performance may drop below their original baseline—because their ability to spot abnormalities independently has weakened. The danger, Nosta argues, isn’t just dependence on AI—it’s regression. As workers become accustomed to AI’s support, their own cognitive abilities atrophy, even as their confidence grows. This creates a dangerous disconnect: people feel more capable, but their actual skill level is declining. This illusion of competence is a growing concern. Nosta points to research showing that AI gives users a false sense of understanding. Rebecca Hinds, head of the Work AI Institute, and Nobel laureate physicist Saul Perlmutter have echoed this, noting that AI can make people believe they comprehend complex topics when they’re simply following algorithmic outputs. In high-stakes environments—such as healthcare, finance, or engineering—this overconfidence can lead to poor decisions, especially when AI is unavailable or fails. The risk is not just inefficiency, but potential harm. Nosta describes the situation as a “cognitive codependent relationship,” particularly among younger workers entering AI-integrated workplaces. While AI can genuinely make people smarter when used intentionally, he cautions that treating it as a substitute for thinking can have the opposite effect—making people dumber over time. Research from Oxford University Press supports this view, finding that AI helps students complete tasks faster but reduces the depth of their thinking. Kimbley Hardcastle, a professor at Northumbria University, goes further, warning of the “atrophy of epistemic vigilance”—the loss of the ability to independently verify, challenge, and build knowledge without algorithmic help. To counter this trend, Nosta advocates for intentional resistance. He urges workers to preserve “cognitive grit,” maintain mental friction, and use AI as a learning tool rather than a shortcut. The goal is not to reject AI, but to use it in ways that strengthen, rather than replace, human thought. In the age of artificial intelligence, Nosta warns, the greatest threat to the workforce may not be machines outperforming humans—but humans forgetting how to think without them. “For the first time in history,” he says, “human cognition is on the obsolescence chopping block.”