Does AI Kill Critical Thinking? Maybe Not If We Use It Right.
Jeanne Beatrix Law's Response to the Recent Microsoft Study's Claims of AI Over-Reliance
Drawing from two years of intensive fieldwork, Educating AI empowers over 4,700 education professionals with battle-tested insights on navigating the AI revolution in education. Our weekly analysis bridges the gap between AI theory and classroom reality, delivering deep systemic perspectives shaped by direct experience with schools and policy development. Your paid subscription demonstrates your commitment to thoughtful education journalism while enabling us to continue this critical work for the entire community.
Nick’s Intro:
For the past week or two, the AI x Education space has been absolutely buzzing about Microsoft's bombshell study "The Impact of Generative AI on Critical Thinking" by Hank Lee et al. The study dives into how knowledge professionals use AI and, through their self-reported experiences viewed through the lens of Bloom's taxonomy, argues that AI reduces users' critical thinking effort. On the surface, this conclusion might seem like stating the obvious – tools designed to simulate critical thinking end up reducing the mental heavy lifting when we use them. Yet here's what's fascinating: many readers (let's be real, probably skimmers) zeroed in on that attention-grabbing subtitle about "Self-Reported Reductions in Cognitive Effort" and are using it as ammunition to challenge the entire project of integrating AI into education.
While I've been wrestling with this study myself, two responses have really lit up my thinking. First, Terry Underwood brings the fire with an extensive critique that dissects the study's design, its reliance on self-reporting, and challenges how transferable its definition of critical thinking really is. You absolutely need to read this one – and please share it far and wide!
The second piece, which I'm thrilled to republish today at Educating AI, comes from the brilliant Jeanne Beatrix Law, Ph.D. As First-Year Writing Director at Kennesaw State University and a fresh voice on Substack, she's doing groundbreaking work at the intersection of AI and university writing instruction. (Dig into her full bio at the article's end.) What makes her take so refreshing is how she coolly acknowledges that yes, AI tools might affect users' critical thinking – but unlike some critics, she doesn't stop there.
Just this week, I read a post from a prominent AI skeptic who weaponized the Microsoft study to double down on their "no productive educational uses of AI exist" stance, calling for a complete freeze on AI training until we find one. But here's the reality: AI is already in our classrooms. Our job isn't to wish it away but to figure out how to harness its inherent properties for our pedagogical goals. And here's the exciting part – many of us are discovering that this shift is absolutely possible through smart implementation, built on shared knowledge from training, courses, and certification processes.
In Dr. Law’s post, you will find another amazing example of this shift!!!
Does AI Kill Critical Thinking? Maybe Not If We Use It Right.
Generative AI is under scrutiny once again. A recent study by Lee et al. (2025)1 argues that AI reduces the effort required for critical thinking, leading knowledge workers to engage less deeply with their tasks. Their survey of 319 professionals suggests that as users become more confident in AI-generated content, they rely on it more and think less critically. In short, AI is supposedly making us lazy thinkers.
That’s a bold and interesting claim—but it also might be an incomplete one.
My research and real-world application of the Rhetorical Prompt Engineering Method (RPM) challenges this narrative.
Instead of AI diminishing our ability to think critically, I’ve seen firsthand how structured prompting enhances metacognition, decision-making, and intellectual engagement. I want to share some preliminary data here that demonstrates my point.
AI and Critical Thinking: The Wrong Question?
Lee et al. suggest that AI shifts cognitive effort from problem-solving to oversight, implying that this shift leads to less engagement with deep thinking. But here’s the problem with that argument:
Critical thinking isn’t just about effort. It’s about strategy.
If AI allows us to automate routine cognitive tasks—like information retrieval or summarization—this doesn’t mean we’re thinking less. It means our thinking is changing. And that shift can be an opportunity rather than a loss—if we learn how to use AI intentionally.
Preliminary Data: The Rhetorical Prompting Model Might Increase Critical Thinking
Let’s move from theory to practice. I recently analyzed feedback from 112 adult learners in Coursera writing courses who used the Rhetorical Prompting Model RPM to guide their AI interactions. These learners possess higher education degrees (28% Bachelor’s); (36% Master’s). More than 56% of them are employed full-time. Their responses tell a different story than the one Lee et al. present.
Preliminary Findings:
✔ 92% strongly agreed or agreed that RPM helped them evaluate their writing choices before and during the writing process.
✔ 75% strongly agreed or agreed that they were able to maintain their authentic voice while using AI assistance.
✔ 89% strongly agreed or agreed that RPM helped them think critically about their writing.
These preliminary numbers make me think: when learners engage with structured prompting, AI doesn’t replace their critical thinking—maybe it amplifies it.
For example, one participant noted:
“Using rhetorical prompts forced me to pause and question why I chose certain words or examples, making me realize how much my audience influences my writing decisions.”
Another learner wrote:
“It was surprising how often I revise on autopilot. Rhetorical prompts helped me understand when a change was necessary versus habitual.”
These responses indicate active cognitive engagement, perhaps not passive AI reliance.
AI Doesn’t Reduce Thinking—It Redirects It
Lee et al. argue that because AI reduces effort in some areas, it leads to less critical engagement overall. But my research shows that AI isn’t eliminating effort—it’s redistributing it toward higher-order thinking.
With structured prompting, learners spend less time struggling with mechanical aspects of writing and more time evaluating, revising, and structuring their work.
How AI Shifts Cognitive Effort:
🧠 From gathering information → To verifying information
🧠 From problem-solving → To integrating AI responses effectively
🧠 From task execution → To overseeing and refining AI-assisted outputs
None of these shifts are inherently bad. They just require a different approach to thinking—one that many traditional models of education haven’t caught up with yet.
AI Literacy: The Missing Piece in the Overreliance Debate
Stewardship, Not Automation, is the Future
Final Thoughts: AI Won’t Replace Thinking—But It Will Reshape It
How do we teach people to prompt better, question better, and think better with AI?
That’s the real challenge—and that’s exactly where my Rhetorical Prompting Model is making an impact, at least based on preliminary results for adult learners. I will continue to update as more students respond and as I scale this work to graduate learners at Kennesaw State.
Dr. Jeanne Beatrix Law is a professor of digital writing, director of the first-year writing, and coordinator of the new graduate certificate in AI Writing Technologies at Kennesaw State University. Her research specialties including multimodal languaging, custom GPTS for creative and applied writing use cases, and generative AI technologies for professional communication. Her public scholarship includes scaling civil rights oral histories into large language models as well as scaling generative for grant writing and everyday use. Jeanne is the co-author of The Writer’s Loop: A Guide to College Writing (Macmillan) and is a founding author for MacMillan Education’s Multimodal Mondays and Bits on Bots blog series.
She has authored chapters on generative AI in edited collections from SIU press, Routledge and the Computers & Composition Press. Her work is also regularly featured in print and digital public media, including The Chronicle of Higher Education. She is the lead researcher for the nationally recognized #ATLStudentMovement digital oral history project and has authored eight courses on Coursera on generative AI use, featuring her Rhetorical Prompting Method (RPM). She is the lead researcher on both external and internal grants to pilot this method. Jeanne also serves as a faculty mentor for the AAC&U’s AI Pedagogy Institute and is on the educator leadership council for Boodle Box and the educator leadership community for OpenAI.
She has been called on as an AI use case expert by private industry and the University System of Georgia on numerous occasions, including serving on a plenary panel and conducting two prompt engineering workshops for the inaugural USG AI Summit. She has presented and published numerous times since 2022 for professional, public, and academic audience on the ethical uses of generative AI, including leading prompt engineering workshops in 2023 and 2024 at Open Educa Berlin (OEB). Her digital and f2f audiences total more than 10,000. Her faculty page for current information: https://facultyweb.kennesaw.edu/jlaw29/index.php and on LinkedIn: https://www.linkedin.com/in/jeanne-beatrix-law-phd-a05b2391/
Check out some of our favorite Substacks:
Mike Kentz’s AI EduPathways: Insights from one of our most insightful, creative, and eloquent AI educators in the business!!!
Terry Underwood’s Learning to Read, Reading to Learn: The most penetrating investigation of the intersections between compositional theory, literacy studies, and AI on the internet!!!
Suzi’s When Life Gives You AI: An cutting-edge exploration of the intersection among computer science, neuroscience, and philosophy
Alejandro Piad Morffis’s Mostly Harmless Ideas: Unmatched investigations into coding, machine learning, computational theory, and practical AI applications
Michael Woudenberg’s Polymathic Being: Polymathic wisdom brought to you every Sunday morning with your first cup of coffee
Rob Nelson’s AI Log: Incredibly deep and insightful essay about AI’s impact on higher ed, society, and culture.
Michael Spencer’s AI Supremacy: The most comprehensive and current analysis of AI news and trends, featuring numerous intriguing guest posts
Daniel Bashir’s The Gradient Podcast: The top interviews with leading AI experts, researchers, developers, and linguists.
Daniel Nest’s Why Try AI?: The most amazing updates on AI tools and techniques
Riccardo Vocca’s The Intelligent Friend: An intriguing examination of the diverse ways AI is transforming our lives and the world around us.
Jason Gulya’s The AI Edventure: An important exploration of cutting edge innovations in AI-responsive curriculum and pedagogy.
 | A guest post by
|
I hear you. What I might add is a bit of my personal pedagogical stance: the working nature of any tool, transformation, strategy, or [insert any noun here] lies somewhere in the middle of two extreme points. What I mean by this is that an integration of generative AI that works for one institutional context may not necessarily work well for another. An example: Anna Mills, a teacher and scholar whose AI work I read and study, was an early advocate for generative AI integration in writing courses. What she found through systematic action research is that her students at her institution needed more formalized support with AI literacy. So, she pivoted towards using AI detection tools in an effort to help her students gain that literacy. My students, matriculating at a large (49K) public R2 institution, don't always need that same degree or specificity of structured support. What works for them, and for adult learners outside of traditional academia, might be something like the prompt-first model. Or, it could be something else. The adaptability and mobility of generative AI seem to support many options for integration. Does that make sense? Thanks for chatting!
This is a great contribution to the conversation in AI and learning. My minor contribution aligns: https://substack.com/@drmountain/note/c-89624068?r=1n866u&utm_medium=ios&utm_source=notes-share-action