GPS ruined our sense of direction. Search engines weaken our memory. AI, scientists warn, could do the same to everything from creativity to critical thinking. Years ago, I deliberately forced myself to start using AI as often as possible. If I was going to be writing about it, I also had to experience the technology firsthand. But an emerging crop of studies over the last year or so have me worried – am I impairing my brain in the process?
These studies suggest people who rely too much on tools like ChatGPT could have problems with creativity, attention span, critical thinking, memory and more. Others raise concerns that AI users could be surrendering the cognitive friction that makes thinking sharp, and that as a society we may have fewer original ideas. But the science on this is brand new, and we don't have the answers. So should we be concerned?
"On a high level, yes," says a professor of neuroscience. There's a lot of nuance here, but AI will do work that used to require mental labour. "There's substantial evidence that if you are not doing as much of the actual thinking, then your capability to do that kind of thinking is going to atrophy." Even if you don't seek to use ChatGPT or Claude, there are AI responses at the top of Google and tech giants are rushing to shove more of it onto our phones. The technology is getting hard to avoid, but there are steps you can take to minimise the biggest potential risks.
This isn't all or nothing though. Using AI doesn't automatically mean it's going to detrimental for you. For example, if AI frees up your brain capacity for other more important things, that might be great for your cognition. "Why do we think AI is going to be that different from other things that our brain has already adapted to? The tools themselves are not inherently good or bad. As with any other technology, how we use AI will determine whether it helps us or harms us. But the concerns are serious enough that you might want to rethink how you use these tools – before it's too late. With this in mind, I spoke with some leading experts in this field to find out how they think we should use AI in ways that preserve and sharpen rather than dull or erode our minds.
Twenty years ago, an idea cropped up that overreliance on technology might cause some kind of "digital dementia" resulting in the deterioration of short-term memory and other cognitive processes. There is currently no evidence for digital dementia. Technology use actually seemed to reduce the risk of cognitive impairment. But that doesn't mean there's nothing to fear.
Studies have found that people who rely on satellite navigation like GPS stop building mental maps of their surroundings and their spatial memory continues to deteriorate over time. A similar phenomenon called the "Google Effect" emerged when search engines took over. Apparently, we're less likely to remember information that we find using a search engine because it takes so little mental effort. It seems the brain gets worse at tasks when we delegate to technology. And AI is the most powerful outsourcing tool of all time.
"What's happening with AI is that it's giving us, for the first time, it gives us an effortless way to bypass the thinking process in favour of polished results." The essay can sound better. The presentation can look sharper. The retirement party joke can land perfectly. But the mental work, the struggle, the false starts and that moment when something finally clicks is exactly what your brain needs. "It's like you're at the gym and a robot lifts the barbell for you," he says. "You get nothing."
So how can you use AI and still give your brain a work out? One recent study found heavier users of AI scored significantly worse on a standard critical thinking test, apparently because they're used to offloading their thought processes to robots. People even trust AI over their own thinking and intuition – even when the AI is wrong – something called "cognitive surrender". The problem gets worse the less you know. A study found you're most at risk when you're less familiar with as subject. "If the user doesn't have the expertise to judge whether the output is great or not. That's the danger."
The fix starts before you open the app. If you wouldn't trust a random stranger's answer to a question, you shouldn't trust AI either. Those are exactly the topics where you need to bring your own judgment first. Try forming a rough viewpoint on the subject and using AI to challenge your perspective before you get the robot's opinion. That way AI is stress-testing your thinking, rather than replacing it.