Frequent AI Use May Be Weakening Our Critical Thinking, Study Suggests

In a digital age where answers are just a prompt away, artificial intelligence has become our ever-present study partner, writing assistant, and problem solver. Yet a growing body of research warns that this reliance may come with an invisible cost: our ability to think critically. A new study published by Michael Gerlich in AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking suggests that people who use AI tools more frequently tend to score lower on measures of critical thinking — the very skill that helps us question, analyze, and reason through complex information.

The research doesn’t claim that AI itself is harmful, but it raises a crucial question about how we’re using it. When technology starts to handle not just our tasks but our thought processes, we risk trading curiosity for convenience. The issue isn’t that AI makes us less intelligent — it’s that over-reliance on it might make us less active thinkers. Understanding why that happens, and how we can prevent it, is key to navigating our partnership with technology in a healthier way.

The Study Behind the Headlines

Gerlich’s research surveyed 666 participants, ranging from teenagers to adults over 60, and asked them about their habits of AI use, their tendency to offload mental work to technology, and their critical thinking performance. The results were striking: those who used AI tools most frequently tended to perform worse on standardized tests of reasoning and analytical evaluation. The link was not direct, however — it ran through what psychologists call cognitive offloading, the process of shifting mental effort onto an external system. The more participants relied on AI to handle intellectual challenges, the less they exercised their own analytical muscles.

This connection echoes long-standing psychological findings about mental “muscle memory.” Just as physical fitness declines without regular exercise, mental acuity wanes when we outsource too many of our cognitive processes. Offloading, in moderation, helps us focus on higher-level thinking — but when it becomes habitual, it subtly discourages the slow, effortful reasoning that critical thinking demands. Gerlich’s findings suggest that young adults, especially those aged 17–25, are most at risk of this effect, as they are both the most frequent AI users and the most comfortable with digital automation. Older participants, who grew up doing more mental work unaided, scored higher in critical reasoning tasks even when they occasionally used AI.

Education level also played a protective role. Participants with higher education maintained stronger critical thinking skills regardless of how often they turned to AI. This implies that formal learning may provide a kind of “cognitive scaffolding,” enabling users to engage with AI critically rather than passively. The difference isn’t in whether someone uses AI, but in how consciously they do it — whether they question, verify, and analyze what the system produces, or simply accept it as correct.

The Psychology of Offloading and Mental Laziness

Cognitive offloading is nothing new. We began outsourcing our memory when we started writing things down, then extended that to calculators, GPS devices, and search engines. What’s different about AI, researchers argue, is its comprehensiveness. It doesn’t just answer factual questions or compute numbers; it now interprets, writes, and creates — domains that previously required deep human thought. As a result, the line between “using a tool” and “thinking for ourselves” becomes increasingly blurred.

Offloading isn’t inherently bad. In fact, it can free cognitive resources for more creative or strategic thinking. The problem arises when offloading becomes habitual and unconscious. When we rely on AI to generate ideas, construct arguments, or summarize information, we may skip the very process of grappling with ambiguity that fosters critical insight. In Gerlich’s study, the more participants described “trusting” AI to handle complex reasoning, the lower their independent critical thinking scores tended to be. The relationship wasn’t absolute, but it revealed a troubling pattern of intellectual passivity: when the machine seems smarter, people stop asking hard questions.

This psychological dynamic is reinforced by convenience. It’s simply easier to let a confident-sounding algorithm do the mental heavy lifting. As neuroscientists have long observed, the brain favors energy efficiency; thinking deeply takes effort. AI’s seductive fluency can short-circuit the natural discomfort that comes from struggling through uncertainty — the very discomfort that teaches us how to reason. As a result, users may mistake well-phrased outputs for well-reasoned ideas, weakening their instinct to critique and verify.

Limitations, Nuance, and What the Data Really Tells Us

The study’s findings are compelling, but like any early-stage research, they deserve careful interpretation. Correlation, after all, is not causation. While AI use was linked with lower critical thinking scores, this doesn’t prove that AI caused the decline. It could be that people with weaker critical thinking skills are simply more inclined to rely on AI, or that certain personality traits—like preference for efficiency—drive both behaviors. The study’s reliance on self-reported data introduces another wrinkle: participants’ own perceptions of how often they use AI or how critically they think may not always reflect reality.

There’s also the question of context. Not all AI use is created equal. Using ChatGPT to summarize a textbook chapter is a very different cognitive task from using it to debate a philosophical question or analyze a dataset. Some forms of engagement with AI may even enhance critical thinking, particularly when users treat AI as a discussion partner rather than an authority. Other studies have shown that when learners are instructed to critique AI-generated responses, their analytical performance improves. In other words, AI can both erode and strengthen critical thinking—depending on how it’s used.

Finally, cultural and educational factors play a role that this study doesn’t fully capture. Access to technology, familiarity with digital tools, and exposure to media literacy education all shape how people interact with AI. A high-school student using ChatGPT to help with essays in a test-driven curriculum may experience different cognitive effects than a researcher using AI to explore new hypotheses. The next step for researchers is to distinguish between passive and active AI use, examining not just how often people use these tools, but how thoughtfully.

The Human Side: When AI Replaces Reflection

Outside of laboratories and surveys, the effects of cognitive offloading are increasingly visible in daily life. Teachers report that students struggle to explain or justify AI-generated work, even when the output is factually correct. In workplaces, managers notice employees submitting polished reports that lack depth or originality—products of a process that skips reflection. And in journalism, AI-assisted writing tools can subtly influence tone and framing, blurring the line between human judgment and machine suggestion.

The Microsoft and Carnegie Mellon study highlighted a similar concern among professionals: 40 percent of tasks completed with AI assistance received little to no critical review. Once people develop confidence in a system’s accuracy, they often stop checking it. This “automation bias” isn’t new—it appeared decades ago in aviation, where pilots overly trusted autopilot systems—but AI brings it into knowledge work. When algorithms offer fluent, plausible answers, our instinct to verify weakens, even when the topic demands scrutiny.

There’s also a cultural cost. The process of thinking deeply, questioning assumptions, and forming original insights is not just a skill—it’s a defining part of being human. When we hand that over to machines, we risk losing not just accuracy but the satisfaction that comes from understanding. As psychologist Daniel Willingham once put it, “Memory is the residue of thought.” When we stop thinking for ourselves, we remember less, learn less, and ultimately, become less curious. The challenge, then, is not to reject AI, but to use it without surrendering the joy and discipline of genuine reasoning.

Rethinking How We Use AI: Building Mental Fitness

If AI can subtly erode our mental sharpness, it can also be harnessed to strengthen it—provided we use it deliberately. The goal is not to stop using AI, but to transform how we interact with it. That begins with awareness. When you ask an AI for help, pause to consider: Am I using this to shortcut the process, or to expand my understanding? A small shift in intent can turn AI from a crutch into a catalyst.

Practical strategies can help preserve our mental agility. Treat AI outputs as drafts, not truths. Always verify claims, challenge assumptions, and look for missing perspectives. Alternating between AI-assisted and AI-free work sessions can also help keep your analytical muscles active—just as cross-training prevents physical overdependence on one exercise. In educational contexts, integrating AI into tasks that require reflection and justification can transform it into a training ground for critical thought rather than a substitute for it.

Long-term, the healthiest relationship with AI may resemble that between a teacher and a student—or perhaps between two colleagues. AI offers speed, scale, and fluency, but humans bring context, values, and judgment. When we use both in concert, we can preserve the essence of what thinking means: the ability to question, connect, and create meaning. AI is a remarkable tool, but it should remain just that—a tool. The work of understanding the world, and ourselves, still belongs to us.