Gloria Mark says AI bots make “depth of processing” disappear, and that’s the problem

SXSW London was loud with music and film, but the real buzz in the room was AI. The sharper question, though, came from Gloria Mark, a psychologist at the University of California, Irvine. She has spent the last 30 years studying how people interact with digital technologies. And in a session titled “Have we lost control of our brains?”, Mark’s answer was blunt: yes.

Mark’s argument connects two dots many people treat as separate. First, she points to her long-running work on attention. Around two decades ago, she began studying how device use affects attention spans by running what she calls “living laboratories,” using sensors and trackers to monitor adult volunteers’ attention, mood, and behavior. In 2003, she found the average user could stay focused on one thing for around two and a half minutes. When she repeated the experiment in 2012, the average had shrunk to around 75 seconds. Then, in research she conducted between 2014 and 2020, attention spans shrank further still to a mere 47 seconds on average.

That’s not just a curiosity. Mark ties rapid switching to stress and performance. In her work, she described having people wear heart rate monitors, and seeing a direct correlation between switching attention fast and stress going up. The fallout extends beyond feelings. “It just takes longer to do any single task if you’re switching your attention,” she said. Mark adds that it’s “not great for performance” and “not great for our emotional well-being.” The implication for executives is simple: if your teams, customers, or audiences keep being trained to hop between stimuli, then deep work gets more expensive. Time management becomes a battle, not a skill.

The AI angle is where her message gets more targeted. In the same conversation, she said large, long-term studies are needed to fully understand how digital technologies affect children. She referenced a current wave of lawsuits and settlements in which Meta, which owns Facebook and Instagram, and Google’s YouTube were ordered to pay millions in damages to a 20-year-old woman who alleged childhood addiction tied to products. She also pointed to a recent settlement involving a rural school district in Kentucky that sought more than $60 million for mental-health needs, with around 1,200 other school districts taking similar legal action against social media companies. The point here is not that a single study proves a single outcome. Mark explicitly says the evidence on social media’s effects on children is inconclusive, despite what best-selling books may claim.

But when she turns to AI chatbots, the mechanism is clearer. Mark argues that a core benefit of human cognition is “depth of processing.” When you evaluate or summarize content, you’re doing what she calls deep processing, which makes you more likely to learn it, understand it, and retain it. With AI bots like ChatGPT, Claude, and Gemini, the pattern changes. When you ask these tools to write, summarize, or evaluate for you, you are no longer doing that depth-of-processing work yourself. “You’re deferring your cognitive work to AI,” Mark said, “and it’s not good for us.” Her concern goes beyond a temporary shortcut. She frames the risk as cognitive atrophy: if you do not constantly “exercise your muscles,” they can weaken over time. In her view, that weakening could make people more vulnerable to misinformation because their critical thinking skills are less practiced.

There is also a relationship layer, and it is not subtle. Mark warns that AI-powered “synthetic companions” can be harmful because human relationships take time, effort, and understanding. If you form a relationship with a sycophantic bot, you skip those skills. The emotional “muscle” at risk is emotional intelligence, which surveys suggest is already on the decline. That is the second-order story boards rarely model: the product can be engaging and still quietly train the user to avoid certain kinds of effort, including effort that builds judgment and empathy.

Mark does not end with optimism theater. She says that if current trends continue, attention spans are diminished, loneliness is rising, boredom is rising, emotional intelligence is decreasing, and studies indicate our sense of purpose is also decreasing. Still, she offers a practical path to course-correct, and it centers on effort. The more effort we put into something, the deeper the satisfaction we stand to gain, she told me. That means choosing to read a book instead of skimming a summary, meeting friends in person when possible, and even trying not to use GPS in places where you can probably manage without it. “I love technology; we can’t give it up,” she said. The line she lands on is about routines: “we have to learn how to create new life routines.”

For decision-makers, the stakes are bigger than wellbeing slogans. Your company either helps users build the kind of mental effort that retains understanding, or it encourages deferral that erodes critical faculties. The same design choice that makes AI feel effortless can also make learning feel optional, and that is how products drift from “assist” to “replace.” The question for boards and operators is whether your product strategy is optimizing for short-term engagement, or for the cognitive and emotional capacity users will need tomorrow, when the novelty wears off and the consequences still arrive.