In a digital age where artificial intelligence is just a click away, it’s no longer a question of whether we can use tools like ChatGPT or Google—but what happens to our brains when we do. A new study from MIT tackles this question head-on, raising concerns that our growing dependence on AI and search engines might be quietly eroding our cognitive abilities. The findings are provocative, suggesting that the more we lean on digital help, the less our brains light up—and that has implications for how we think, learn, and remember.
The Experiment: Brains vs. Bots
The study, conducted by researchers at MIT’s Department of Brain and Cognitive Sciences, compared brain activity in individuals performing complex reasoning tasks under three different conditions:
- Brain-Only – No external assistance allowed.
- Google Help – Participants could search online.
- LLM Help – Participants could use a large language model like ChatGPT.
Functional MRI (fMRI) scans revealed something striking. Participants in the Brain-Only condition showed significantly stronger and more widespread neural activity, especially in regions linked to working memory, abstract reasoning, and executive control. Those using Google or LLMs displayed reduced connectivity, indicating that the brain was outsourcing a substantial portion of the mental workload.
What Happens to Your Brain When You Use an LLM?
Using a language model like ChatGPT simplifies information retrieval and synthesis. But the ease comes at a cognitive cost. Instead of working through problems or constructing arguments from memory, users tend to passively accept and process AI-generated responses. According to the study, this shift corresponds with a measurable drop in brain network engagement—particularly in the prefrontal cortex, which plays a key role in decision-making and critical thinking.
It’s not that LLMs erase your brainpower—but they restructure how your brain is engaged. The more users rely on AI, the less the brain needs to “struggle”—and struggling, in this context, is a vital part of learning.
The Memory Meltdown
The MIT team also looked at the effects on memory retention. Unsurprisingly, participants in the Brain-Only group retained information significantly better than those using AI or search engines. This ties into the concept of “cognitive effort”—when we exert more mental energy to encode information, we remember it more deeply.
This isn’t new: previous studies on “Google Effect” or “digital amnesia” showed similar trends. But the new data from LLM use suggests the impact may be even more pervasive. With LLMs generating complete answers—often with contextual nuance and fluency—it becomes tempting to bypass the mental labor altogether.
What Happens When the AI is Taken Away?
Here’s where things get even more interesting.
Participants who had previously relied on Google or ChatGPT performed worse when the tools were removed. Not only did they struggle to complete tasks independently, but their brains also showed weaker activity patterns compared to the Brain-Only group—even after adjusting for prior ability.
This suggests that AI assistance might not only reduce immediate cognitive effort, but also impair the brain’s ability to “bounce back” when help is no longer available. It's like cognitive atrophy—the longer we lean on the crutch, the harder it becomes to walk without it.
Are We Accumulating a “Cognitive Debt”?
The researchers introduced a compelling concept: cognitive debt. Just as financial debt accumulates when you borrow against future income, cognitive debt builds when you consistently offload thinking to machines.
At first, it feels efficient—tasks are easier, faster, and less mentally taxing. But over time, the hidden cost is a weakening of the mental muscles we rely on for critical thinking, problem-solving, and creativity. The more we automate our thinking, the more we risk becoming dependent on the very tools we built to empower us.
So, Is ChatGPT Making Us Dumber?
Not exactly. The study doesn't claim that ChatGPT—or AI in general—causes stupidity. Rather, it suggests that habitual reliance on these tools diminishes cognitive engagement. When we offload too much mental work, we use our brains less actively and deeply, which can impair learning, memory, and resilience.
In other words, it’s not the tool that makes us dumber—it’s how we use it.
The MIT study serves as a timely wake-up call. As AI tools like ChatGPT become integrated into every corner of our lives—from classrooms to boardrooms—we must rethink the balance between efficiency and engagement.
AI can be an incredible partner in learning and creativity, but only if we remain active participants in the process. Instead of treating LLMs as cognitive replacements, we should treat them as cognitive scaffolds—a temporary support while we build and strengthen our own mental abilities.
In the end, the choice isn't between brains and bots. It’s about making sure our brains don’t go silent in the presence of bots.