Step into the Massachusetts Institute of Technology (MIT) Media Lab in Cambridge, US, and the future feels a little closer. Glass cabinets display prototypes of weird and wonderful creations, from tiny desktop robots to a surrealist sculpture created by an AI model prompted to design a tea set made from body parts. In the lobby, an AI waste-sorting assistant named Oscar can tell you where to put your used coffee cup. Five floors up, research scientist Nataliya Kosmyna has been working on wearable brain-computer interfaces she hopes will one day enable people who cannot speak, due to neurodegenerative diseases such as amyotrophic lateral sclerosis, to communicate using their minds.

Kosmyna spends a lot of her time reading and analysing people’s brain states. Another project she is working on is a wearable device – one prototype looks like a pair of glasses – that can tell when someone is getting confused or losing focus. Around two years ago, she began receiving out-of-the blue emails from strangers who reported that they had started using large language models such as ChatGPT and felt their brain had changed as a result. Their memories didn’t seem as good – was that even possible, they asked her? Kosmyna herself had been struck by how quickly people had already begun to rely on generative AI. She noticed colleagues using ChatGPT at work, and the applications she received from researchers hoping to join her team started to look different. Their emails were longer and more formal and, sometimes, when she interviewed candidates on Zoom, she noticed they kept pausing before responding and looking off to the side – were they getting AI to help them, she wondered, shocked. And if they were using AI, how much did they even understand of the answers they were giving?

With some MIT colleagues, Kosmyna set up an experiment that used an electroencephalogram to monitor people’s brain activity while they wrote essays, either with no digital assistance, or with the help of an internet search engine, or ChatGPT. She found that the more external help participants had, the lower their level of brain connectivity, so those who used ChatGPT to write showed significantly less activity in the brain networks associated with cognitive processing, attention and creativity.

In other words, whatever the people using ChatGPT felt was going on inside their brains, the scans showed there wasn’t much happening up there.

The study’s participants, who were all enrolled at MIT or nearby universities, were asked, right after they had handed in their work, if they could recall what they had written. “Barely anyone in the ChatGPT group could give a quote,” Kosmyna says. “That was concerning, because you just wrote it and you do not remember anything.”

Kosmyna is 35, trendily dressed in a blue shirt dress and a big, multicoloured necklace, and she speaks faster than most people can think. As she observes, writing an essay requires skills that are important in our wider lives: the ability to synthesise information, consider competing perspectives and construct an argument. You use these skills in everyday conversations. “How are you going to deal with that? Are you going to be, like, ‘Err … can I just check my phone?’” she says.

The experiment was small (54 participants) and has not yet been peer reviewed. In June, however, Kosmyna posted it online, thinking other researchers might find it interesting, and then she went about her day, unaware that she had just created an international media frenzy.

Alongside the journalist requests, she received more than 4,000 emails from around the world, many from stressed-out teachers who feel their students aren’t learning properly because they are using ChatGPT to do their homework. They worry AI is creating a generation who can produce passable work but don’t have any usable knowledge or understanding of the material.

The fundamental issue, Kosmyna says, is that as soon as a technology becomes available that makes our lives easier, we’re evolutionarily primed to use it. “Our brains love shortcuts, it’s in our nature. But your brain needs friction to learn. It needs to have a challenge.”

If brains need friction but also instinctively avoid it, it’s interesting that the promise of technology has been to create a “frictionless” user experience, to ensure that, provided we slide from app to app or screen to screen, we will meet no resistance. The frictionless user experience is why we unthinkingly offload ever more information and work to our digital devices; it’s why internet rabbit holes are so easy to fall down and so hard to climb out of; it’s why generative AI has already integrated itself so completely into most people’s lives.

We know, from our collective experience, that once you become accustomed to the hyperefficient cybersphere, the friction-filled real world feels harder to deal with. So you avoid phone calls, use self-checkouts, order everything from an app; you reach for your phone to do the maths sum you could do in your head, to check a fact before you have to dredge it up from memory, to input your destination on Google maps and travel from A to B on autopilot. Maybe you stop reading books because maintaining that kind of focus feels like friction; maybe you dream of owning a self-driving car. Is this the dawn of what the writer and education expert Daisy Christodoulou calls a “stupidogenic society”, a parallel to an obesogenic society, in which it is easy to become stupid because machines can think for you?