We Don't Have to Lose Critical Thinking with AI

AI reducing critical thinking isn’t inevitable, it depends on our choices and changing how we teach

A colleague at the McCombs School of Business forwarded to me a recent Wall Street Journal article, "How I Realized AI Was Making Me Stupid—and What I Do Now," stating he had similar concerns that increasing reliance on AI could potentially hinder the development of essential critical thinking skills.

Schechner, Sam. 2025. How I Realized AI Was Making Me Stupid—and What I Do Now. The Wall Street Journal, April 3, 2025.

There are valid concerns about the impact of AI on critical thinking. Employment, innovation, and engagement in civic life will suffer if we graduate students ill-equipped to analyze information, discern truth from misinformation, solve complex problems, and make reasoned judgments. Yet, this potential negative outcome is not inevitable. Recent studies suggest that AI use can correlate with reduced critical thinking due to "cognitive offloading," particularly among younger users. However, other research indicates AI can enhance learning, foster deeper insights, and free up cognitive resources for higher-order analysis when used appropriately. For instance, time saved on routine data summarization could be redirected towards critically evaluating the underlying assumptions of different business strategies or synthesizing conflicting market reports into a more nuanced forecast. Maintaining critical thinking requires intentional strategies for both individuals in how they utilize AI and educators in how we train our students to verify AI output and use AI as a partner, redesigning assignments to promote rigor and synthesis. Effectively harnessing AI requires a shift in how we structure our courses and train our students to use these powerful AI tools not as shortcuts but as catalysts for deeper learning and sharper critical thinking.

Why Worry About Critical Thinking?

Critical thinking is analyzing information objectively, evaluating arguments logically, and making reasoned judgments, what my mother called “thinking for myself.” It involves questioning assumptions, identifying biases, considering different perspectives, and drawing well-supported conclusions rather than simply accepting information at face value.

This skill is crucial. It empowers individuals to solve problems effectively, make sound decisions in their personal and professional lives, navigate the complexities of misinformation in the modern world, and contribute thoughtfully to society as informed and engaged members. Without critical thinking, we are more susceptible to manipulation, poor reasoning, and an inability to adapt to new challenges.

Emerging Concerns About Critical Thinking

The WSJ article highlights two recent studies that have fueled concerns about AI and human thinking. A recent UK study has garnered attention for revealing a negative correlation between the frequent use of AI tools and critical thinking skills. The researchers attribute this to increased "cognitive offloading"—delegating thinking and problem-solving—which reduces the ability to think independently. They note, "This relationship underscores the dual-edged nature of AI technology; while it enhances efficiency and convenience, it inadvertently fosters dependence, which can compromise critical thinking skills over time." The study found that younger participants (17-25) showed higher dependence and lower critical thinking scores. However, the study also offered a counterpoint: higher educational attainment correlated with better critical thinking skills and appeared to moderate AI's negative impact. With education, we can equip students to use AI critically.

Similarly, a recent Microsoft research survey found that while AI reduces the perceived cognitive effort in knowledge work, it risks diminishing critical thinking through over-reliance. This risk appears heightened when users lack domain expertise or place excessive confidence in the AI, shifting their role from active thinking to mere verification.

Historical Parallels and The "Google Effect"

These concerns about the cognitive impact of technology echo throughout history. Millennia ago, in “Phaedrus,” Plato records Socrates famously worrying that the advent of writing would atrophy memory as people relied on external symbols instead of internal recollection. More recently, 2011 research on the "Google effect" found that instead of memorizing information, people increasingly remembered where to find it online, effectively treating the internet as an external hard drive.

However, I differentiate between specific task competencies that technology might render obsolete, like navigating solely by paper map when most people will only ever use GPS, and foundational cognitive skills such as critical thinking. While our tools evolve, the underlying need for critical analysis, logical evaluation, and reasoned judgment persists and arguably becomes even more vital. We should expect a continuation, and perhaps acceleration, of cognitive offloading with AI. As instructors, our role in cultivating critical thinking will become even more important.

Evidence AI Can Enhance Critical Thinking

However, our understanding of the impact of AI on human thinking is evolving, and there is evidence that AI can be a catalyst for critical thought if used correctly. A 2024 study in Indonesia focusing on English Education majors found that AI use assisted in expanding ideas and providing deeper insights. However, its effectiveness depended on the user’s ability to ask precise questions and interpret the outputs. Another 2024 study involving 200 students reported a 10% increase in critical thinking skills, such as problem-solving and independent thought, after integrating AI tools with high student engagement. By automating routine tasks, AI can free up cognitive bandwidth for more valuable higher-order analysis and creative problem-solving provided users engage with AI as a partner, not just a shortcut.

A Depth vs. Breadth Tradeoff

AI offers an incredible breadth, quickly surveying vast amounts of information, while traditional close reading provides depth crucial for critical thinking. Managing this tradeoff requires informed choices. Two of my most common uses for AI are prompts to summarize an academic paper, or summarize a reading, write learning objectives, and create discussion questions. (You can try my prompts for yourself.) I use AI to map the content and understand key points (breadth) before deciding whether I need to read for deeper understanding. If I decide that a paper is sufficiently essential or complex, or if it is a reading for a class I am teaching, I read it thoroughly, evaluate the evidence, and form my conclusions (depth). Sometimes, I use a hybrid approach, reading an AI summary and then chatting with the AI about the details of the paper to drill down on specific concepts or implications. Productivity and learning result from strategically choosing when AI assists with breadth and when to engage directly for depth, harnessing AI without sacrificing critical understanding.

Strategies for Fostering Critical Thinking with AI

How can we reap AI's benefits while strengthening critical thinking? Research and practical experience suggest several approaches for both individuals and educators.

For Everyone: Using AI as a Partner

• Trust but Verify: Never blindly accept AI output. Treat it as a first draft requiring scrutiny for accuracy, bias, and relevance. In my courses, I emphasize "trust but verify." Students must edit AI output to understand it and make it their own. Build prompts to critique AI responses into assignments and workflows. Evaluate an AI-generated competitive analysis: Are the identified competitors truly relevant? What market segments did AI miss?

• Feed Curiosity: Use the time saved by AI on routine tasks to engage in deeper analysis, strategic thinking, and questioning. Reflect on how you reach conclusions, whether with or without the aid of AI. Understand the reasoning, not just the result. After AI generates sales forecast data points, spend the saved time exploring why certain trends are projected and probing for underlying market drivers.

• Ask Probing Questions: Don't be a passive recipient. Challenge assumptions in AI responses, ask for alternative perspectives, request evidence for its claims, and explore edge cases. If AI suggests a specific marketing channel, ask, "What data supports this channel's effectiveness for our target demographic compared to others?" or "What are the potential downsides or risks of focusing heavily on this channel?"

• Make Your Own Decisions: Employ AI for brainstorming, initial research, or structuring routine tasks (like replying to an email). However, ensure you perform the core analysis, strategic decision-making, and critical evaluation yourself. Ask AI to brainstorm potential risks for a new product launch, then use your own judgment to prioritize, analyze, and develop mitigation strategies for the most critical risks.

For Instructors: Redesign to Improve Learning

• Design Multi-layered Assignments: Expand beyond tasks AI can easily complete. Create assignments requiring students to connect concepts across different readings, case studies, class discussions, and personal experiences, forcing synthesis and independent conclusion-forming. Evaluate assignments on the use of terms and concepts from class discussions and readings, not general concepts generated by AI. Ask students analyzing a company's financial distress to integrate concepts from finance (ratio analysis), strategy (industry analysis), and potentially organizational behavior (leadership issues). Evaluate on use of course terms and concepts.

• Increase Rigor: If a traditional assignment involved applying one framework (e.g., SWOT analysis), evolve it. Require comparing multiple frameworks, justifying the choice, or applying a framework to a novel, ambiguous situation demanding more judgment than AI typically offers. Instead of just identifying Porter's Five Forces, ask students to evaluate which force is most impactful for a specific startup and justify why, connecting it to the company's unique value proposition.

• Integrate More Discussion: Written assignments are susceptible to direct AI generation. Shift emphasis towards activities requiring real-time synthesis and articulation. Use assignments that culminate in class debates, presentations, or simulations where students must defend their reasoning and respond dynamically. Instead of having students submit a report on business ethics, have them write a memo that they propose to a group of peers who evaluate the quality of their argument and responses to questions.

Looking ahead, AI design itself might evolve to support critical thinking. AI interfaces could be designed not just as assistants but as "provocateurs." These AI systems would intentionally introduce critiques, highlight potential biases, or suggest alternative viewpoints alongside their primary output. The goal is to actively stimulate user reflection and critical engagement, creating tools that build, rather than erode, cognitive skills.

Cultivating Critical Minds in the Age of AI

The integration of generative AI into education and professional life is inevitable. While concerns about its impact on critical thinking are valid, whether we lose critical thinking depends on our choices. Succumbing to fear or outright bans is unproductive and would lead to other losses. Recent evidence suggests a more nuanced reality: the impact of AI depends heavily on how it's used and how we teach it. As educators, our role is crucial in guiding this integration. By embracing AI thoughtfully, redesigning our teaching strategies to emphasize synthesis, evaluation, and independent reasoning, and training students to interact with AI as critical collaborators, we can mitigate the risks. We can help our students leverage AI to free up cognitive resources, enabling them to tackle more complex problems and ultimately enhance their critical thinking capabilities, preparing them to thrive personally and professionally as innovators and responsible citizens.

Thank you to Derrick Bonyuet, Clinical Assistant Professor in the Department of Accounting at the University of Texas at Austin, McCombs School of Business, for his suggestions on this post and to Amitai Bernstein for her comments and editing.