The Researcher Who Studies Attention Says AI Chatbots Are Hollowing Out How We Think

Speaking at SXSW London, UC Irvine professor Gloria Mark — who has spent 30 years measuring how digital tools fragment human attention — told the audience that AI chatbots represent a qualitatively different problem than smartphones or social media. Reporting from MIT Technology Review captures her central argument: the danger is not that AI makes people less intelligent, but that routinely delegating thinking to a language model degrades the metacognitive skills required to know when your own reasoning should be trusted.

Mark's research has tracked how each generation of digital tools — email, social media, mobile notifications — shifts cognitive load in ways that take years to show up in measurable outcomes. LLMs, she argues, accelerate that process by taking over tasks that once required sustained independent effort: summarizing, structuring an argument, drafting a response, deciding what to do next. The work still gets done. The muscle that used to do it goes unused.

The timing of the warning matters. Schools are normalizing AI tool use in classrooms at speed. Corporations are mandating it across workflows. The pace of adoption leaves almost no space for the longitudinal research that would document what is actually being lost. By the time solid data exists, the habits will be deeply entrenched.

Every generation claims the new technology is eroding young minds, and every generation is partially right and mostly wrong. Mark's argument is more precise than a moral panic — which is exactly what makes it worth engaging seriously.

Read the full story at MIT Technology Review