Is AI Making You Less Skilled? The Hidden Cost Of Letting Machines Think For You

AI deskilling is quietly eroding our ability to write, code, think, and make decisions—and most of us don't even notice. As we hand over more cognitive work to artificial intelligence, the very skills that make us competent professionals are atrophying, warns Forbes contributor Bernard Marr in a provocative new analysis.

At the core of the argument is a simple paradox: the more we use AI to boost productivity, the less we practice the fundamental skills that underpin expertise. Writers lean on generative models for drafts, coders rely on AI assistants for debugging, and managers trust algorithms for strategic decisions. Each shortcut saves time in the moment—but at a hidden, long-term cost.

This isn't a new concern. The concept of "deskilling" dates back to industrial automation, where machines replaced manual craftsmanship. What's different today is the scale and speed. AI doesn't just replace physical labor; it replaces cognitive labor—thinking itself. Research from Microsoft and Carnegie Mellon in 2023 suggested that heavy users of AI exhibit measurably lower critical thinking scores. A more recent study by the University of Oxford (2025) found that programmers who rely on AI coding tools struggle more when asked to debug without assistance.

Marr identifies three domains most at risk: writing, coding, and decision-making. In writing, AI-generated prose reduces the need for structure, argument, and voice. Coders become assemblers rather than engineers, pasting snippets without understanding underlying logic. Decision-makers accept AI recommendations without evaluating assumptions or biases. The result is a workforce that is faster but shallower—operationally efficient but strategically brittle.

The hidden cost of AI deskilling extends beyond the individual. Organizations face a growing gap between their reliance on AI and the internal capacity to challenge or improve its outputs. When a machine suggests a flawed answer, it's the human who must catch it—but if that human has lost the skill to detect error, the entire system becomes fragile. This echoes what aviation experts call "automation bias": pilots who trust autopilot so much that they fail to intervene when it fails.

Looking ahead, the antidote is not to abandon AI but to use it deliberately. Experts recommend structured training that pairs AI tools with manual practice, regular "no-AI" exercises, and a culture that rewards questioning rather than accepting outputs. As Marr concludes, the goal should be augmentation, not replacement—but that requires conscious effort. The quiet erosion of skill is happening now; whether we reverse it or accelerate it depends on how we choose to work with machines.