Not exactly—but how we use it matters.

A new Trends in Cognitive Sciences perspective argues that AI doesn’t inherently erode human intelligence. Instead, it highlights a well-known principle in cognitive psychology: cognitive offloading.

When we let AI perform tasks that require reasoning, writing, memory, or problem-solving, we reduce the amount of mental practice our brains receive. Like physical exercise, cognitive skills strengthen through use and weaken through disuse.

Skills: learned abilities such as writing, mathematical reasoning, diagnosis, or programming. These are most vulnerable if AI consistently replaces the learning process.

Basic cognitive abilities: foundational functions like working memory, attention, and executive control. Current evidence suggests these may be more resistant to decline, although more research is needed.

The key message isn’t that AI makes people “stupid.” Rather: AI can improve immediate performance. Overreliance may reduce long-term learning and skill retention.

AI is most beneficial when it augments human thinking instead of replacing it. This fits with decades of neuroscience showing that practice drives neuroplasticity. The brain adapts to the cognitive demands we place on it. If.

AI removes those demands entirely, we may become more efficient today while sacrificing expertise tomorrow. The question, then, isn’t “Should we use AI?” It’s “How can we use AI to enhance learning without outsourcing the very processes that build intelligence?”

Reference: Cash TN, Kelly MO, Macnamara BN, & Risko EF. Is AI making us stupid? Trends in Cognitive Sciences (2026).

Many worry that AI use will erode human cognition. We discuss evidence suggesting that offloading cognition to AI can impede skill acquisition and lead to skill decay, but risks depend on how AI is used. We also consider whether our basic cognitive abilities might prove more resilient to this erosion.