For centuries, technology has helped us think more efficiently. Calculators took over arithmetic. GPS replaced navigation. Each innovation reduced mental effort while leaving the business of thinking largely untouched.
Artificial Intelligence feels different. It isn’t memory or calculation being outsourced anymore. It’s brainstorming, drafting, analysing and deciding. Thinking itself is becoming something that can be delegated. In that quiet handover, a new kind of exhaustion is taking root.
Welcome to AI burnout.
Clients arrive with anxiety they cannot quite locate, a compulsive need to second-guess themselves, and a persistent sense of professional inadequacy. Recent research has found that heavy AI use is associated with lower critical thinking through cognitive offloading, relying on external tools instead of engaging in deep, reflective thought. Used in moderation, this is efficient. In excess, it quietly weakens our willingness to tolerate uncertainty or wrestle with difficult ideas ourselves.
Confidence begins to migrate from our own judgment to a machine’s output. Psychologists call this erosion of self-efficacy: the gradual loss of belief that you can solve problems on your own. Clients also describe work that no longer feels like theirs. The words are technically their own, yet somehow unfamiliar. AI creates a peculiar kind of self-doubt: the feeling of being an imposter despite having completed the work yourself.
AI also changes the nature of mental effort. Rather than removing judgement, it often multiplies it. Multiple polished answers still require someone to evaluate, compare and decide. The cognitive load hasn’t disappeared; it has simply changed shape while stripping away the satisfaction of arriving at an answer yourself. These effects don’t stop at the workplace. They are beginning to appear in classrooms too.
School was never simply about producing the right answer. It was about learning to tolerate uncertainty, make mistakes, and reason your way towards understanding. That struggle creates the cognitive friction essential for learning and, with it, genuine confidence. AI can bypass that process entirely. Students may submit polished assignments without developing the patience, curiosity or persistence that real learning demands.
None of this makes AI the villain. Like any cognitive tool, its psychological impact depends on how we use it. The question isn’t whether AI should think with us. It’s whether it is beginning to think instead of us.
Perhaps what we now need is AI hygiene. Attempt the first draft before opening the app. Stay with the difficult problem a little longer. Use AI to challenge ideas you’ve already formed rather than forming them for you. The discomfort of not immediately knowing the answer isn’t a flaw in your thinking. More often than not, it’s where thinking begins.
AI can generate ideas and mimic expertise. What it cannot build for us is judgement. That remains irreducibly human. But only if we continue to practise it.