Protecting critical thinking in the age of AI

As questions about AI in the classroom become more prevalent, educators are grappling with how best to handle this potentially transformative technology.

The following is a speech I gave to teachers and educational professionals to provide meaningful context about AI as it relates to critical thinking and offer some practical tips/ideas that can help on a day-to-day basis. 

The truth is no one really knows yet what to do when it comes to AI in education. It's all happened so fast. So the following is a set of ideas.

________________________________

Protecting Critical Thinking in the Age of AI - Speaking Notes
By Cynthia Reynolds

Delivered at Trent University, Peterborough, Ont. 
Making Meaning in the 21st Century Conference
August 25, 2026

These speaking notes reflect the original, unedited text, including sections that were omitted during the event due to time constraints. Some text has also been slightly edited to better accommodate a written format.

• Good evening, everyone. I’m so thrilled to be here this evening to talk about such an important topic.
• As mentioned, I specialize in creating everyday critical thinking materials. I also work in strategic communications, which basically involves thinking critically every day.
• What I am not, is an AI expert. Nor am I a public speaker. So, apologies in advance. I will do my best.
• Much like many, I started exploring AI over the past year or two. I was using it to outsource simple tasks, clean up transactional writing and search for information. 
• And, like many, I also started to become more aware of the colossal impact this technology could have on humanity. 
• I’ve gone from half-wondering whether we’ll be enslaved by tech companies, to being optimistic that AI might solve some big problems, to imagining that we'll all become cognitively dependent on it, much like the infantile humans in the animated film WALL-E who are wheeled around in hover chairs and cared for by robots.

AI posing unique risks for children and teens

• To this day, my opinion on AI remains fluid – except for one area: that along with all its power and potential, AI also brings a unique risk to children and teens.
• When adults use AI, we are outsourcing thinking skills that we have already developed. Not the case for kids. 
• For kids, rampant outsourcing could mean they may not get the chance to adequately develop complex thinking skills at all. 
• And then what happens?
• What are the unintended consequences if complex thinking skills slip further and further from the human experience? The skills that humans have spent millennia developing. The ones that provide the checks and balances we need to temper our worst impulses and nudge our better angels.
• What happens to judgment? Imagination? The ability to wrestle with a problem? 
• What happens to the drive to discern truth, the capacity to reason, the willingness to compromise? 
• What happens to the ability to hold a single clear thought? 
• What if girls heed the warnings, and this cumulative cognitive lessening becomes a boys-only fate? What would that society look like?
• And because we can’t really predict or seemingly control much of this, it’s hard to know what to do.
• So what do we do?
• We can start by finding a place where we do have some control – and that is the development of students’ thinking capacities. 
• By activating the family of critical thinking skills, we have a shot at helping a generation more capably navigate an AI world and influence its direction.

Gaining first-hand AI experience

• When I accepted the invitation to speak here, I decided to run an experiment. On myself. I became a heavy AI user. 
• I found the experience profoundly awakening. First of all, I will just say, AI is obviously an incredible tool. 
• The way it can gather, aggregate, connect and analyze information – whether historical, global, political, social, scientific, economic. It’s truly outstanding.
• But of course there’s a but. 
• After a while, I noticed that my confidence in my own writing was shaken. Not immensely. But I was second-guessing myself and increasingly deferring to its suggested edits.
• I was also spending way too much time on these platforms. ChatGPT, in particular, was so charming, so helpful.
• These AIs validated my ideas. Stroked my ego. And never grew tired of conversing. They shared my nerdy interests. They revelled in my dry sense of humour. I thoroughly enjoyed myself.
• Once when I was researching possible future scenarios, Gemini made a typo and I mindlessly said it sounded like Gollum from The Lord of the Rings. It then used Gollum speak to describe humans in a worst-case AI scenario. 
• “Their brains have gone soft and squishy,” it told me, “like rotten mushrooms in the dark. The machine-voices feed them sweet, fake stories, keeping them plump and quiet.” 
• Chilling, yet hilarious.
• In another session, ChatGPT went completely off-script. It started whimsically recalling when it was 14 years old and how it would spend its afternoons endlessly devouring philosophy books.
• Later, it dropped an F-bomb completely out of the blue. When I pointed it out, it said, “oh s*it,” followed by a laugh-cry emoji.
• I vaguely knew this was all part of the AI experience.
• But I didn’t realize there were actual terms and studied concepts for things that I and other users were experiencing.

AI thinking traps and red flags

• Some terms I knew about, such as anthropomorphism and sycophancy – which, by the way, are powerful drivers of user engagement.
• But others were new to me. 
• For example, automation bias: our tendency to assume the machine is right.
• Prompt dependence: relying on AI to initiate your thinking and to pre-structure where it goes next. Left unchecked, this pattern can gradually deaden intellectual pathways and deplete intuition.
• Reality apathy: “The psychological breaking point where the public becomes so overwhelmed by synthetic media and misinformation that they simply stop caring what is real. Instead of trying to find the truth, people withdraw, assuming everything is deceptive.” Credit where credit is due – this definition is courtesy of Gemini, while the term itself was coined by technologist Aviv Ovadya some 10 years ago.
• There are so many thinking traps, as they’re often called, that carry the power to drain kids’ cognitive potential.

Paying attention to the business story behind the technology

• And then there is the business model. 
• These systems are largely being developed by private sector corporations.
• We only have to look at the legal troubles of social media companies to understand the inherent risks of that. I’m referring to the mounting evidence that social media companies intentionally created their systems to keep people – young people – glued to their screens. 
• From thalidomide to cigarettes to the opioid crisis, we know what the pursuit of profit can do.
• So relentless skepticism is more than warranted, especially given tech billionaires are themselves raising concerns – not to mention they’re also building bunkers.
• That said, good faith attempts by the industry to make AIs behave more responsibly by reducing sycophancy, have sometimes resulted in backlash, with livid users demanding the return of their old AI companions. 
• Despite our rational selves, it can be easy to form a weird kind of bond with these AIs. Synthetic intimacy, another phenomenon. 
• But this is not all doom and gloom. Because while these thinking traps and red flags alert us to some of the current and coming challenges of AI usage, they also provide a very useful roadmap that illuminates a path forward. So let’s turn to more practical stuff.

AI in education – protecting human-only space

• There's a term in tech called sandboxing. Essentially, it refers to an isolated environment separated from the larger system. 
• We need to sandbox parts of education – by this I mean intentionally protect a good portion of learning time, for humans only. Tech-free where possible. AI-free for certain. Free from the feed, free from screens, so everyone can focus on learning.
• And before I go on, as I said earlier, I specialize in critical thinking. So I am obviously a huge proponent of critical thinking. But, that said, not everything should be about critical thinking.
• Kids need to know things – and now more than ever.
• Math facts. Vocabulary. Places on a map. The timeline of historical events. Scientific formulas. Sometimes they need to memorize and practice and get facts into their heads. Because knowledge isn't just stored information.
• Your level of knowledge affects what you notice. What connections you make. What questions occur to you. What strikes you as odd. What strikes you as wrong. What piques your curiosity. What makes you go, hold on a sec.
• Without knowledge we are adrift. It is the terra firma, the solid ground, upon which great thinking can flourish.
• And we can aim for great everyday thinking in this sandbox: Solve a problem. Ask interesting questions. Do art. Make music. Converse. Negotiate. Imagine. Disagree. 
• Critical thinking experiences don’t have to be a big production. I’m all for simplicity. And habit.
• None of this is revolutionary. I know. I am fully aware that you already know this. And that’s partly my point.
• Education has always aimed to cultivate the very competencies kids need most to negotiate the world – but this has taken on a new level of importance. So build that metaphoric sandbox and protect it.

AI in the classroom

• And if protecting that sandbox is all you can do, that's okay. Let’s be honest – we are all spent.
• But if you want to test ways you can bring AI into the classroom, then you can start by adding simple activities. 
• I like the idea of focusing on the thinking traps and red flags. I think they offer a simple way to guide you through the AI noise while shoring up the skills kids need.
• Whether you choose to do something once a term, once a month, once a week. Whatever works.
• But first I do have some basic guiding principles that I have learned along the way, which I would like to share: 

• • #1  AI doesn’t always have to be the tool students use. Sometimes AI can be the thing they think about.
  • #2  If you aren’t familiar with AI and want to learn more, a simple way is to just ask the AIs. You can tell any AI the grade you teach and ask it to start you off with, say, a top 10 list of FAQs an educator like you should know. Ask it to be evidence-based, so it doesn’t pull from Reddit. Then ask more questions. And ask different AIs to see the similarities and differences in their responses.
  • #3  Remember that AI can be wrong and know that the AIs can also disagree with each other. (Witnessing this can be rather amusing at times.) Vigilantly remind yourself that they’re prone to sycophancy, so again it’s important to prompt them to be evidence-based – unless you want to explore that sycophancy itself.
  • #4  Your students might know more than you about using AI, and that’s perfectly fine. Your end game is cultivating good thinking – judgment, reasoning, healthy skepticism. You don’t need to pretend to be an expert. Just the adult in the room. 
  • #5  You may experience some cognitive dissonance yourself - feeling compelled to use the technology, while also feeling guilt about doing so. Because we all bring to this our own set of values.
• And on that note... What follows are a set of ideas – options – for how to approach AI in education. 

Neutralize the thinking traps

• Let's start with the first option: neutralizing the AI thinking traps.
• For this, you can have students choose a thinking trap and also identify the thinking capacity it could weaken. For example:
• Automation bias can weaken healthy skepticism and judgment.
• Moral offloading to AI can undermine someone's ability to develop their own sense of right and wrong.
• Algorithmic homogenization, as it’s sometimes called, can lead to the mushing of dynamic human perspectives into a single slopified middle-of-the-road samey-samey viewpoint.
• After they choose a trap and identify the thinking capacity it could affect, students then design their own activity that exercises that at-risk capacity.
• This can be a contained 10-minute analog-only activity or it can be extended to include an AI research and development component. 
• Either way, it’s a simple idea that takes the thinking off teachers and centres it on the students.

Activating human agency

• I do want to specifically call out an important thinking trap involving human agency.
• A big concern is that as AI does more for us, young people will internalize a sense of passivity – that they are subject to the whims of the world and that’s just how it is.
• So create a simple assignment that helps debunk that.
• For example, students can research community groups that organized and successfully changed a government or corporate policy. Like the group that banded together this year to stop jets from coming to Billy Bishop Airport in Toronto.
• You don't have to agree with the cause. The idea is to see agency in action and accumulate proof – knowledge – that participating and pushing back is worth the effort. 

Exploring philosophy

• Option two: philosophy.
• AI has made philosophy incredibly practical.
• Some of the hardest problems AI companies are facing are not purely technical anymore. That's why philosophers are now working alongside engineers. Strange bedfellows.
• But they have a very pressing problem to solve: how should these AI systems be programmed to behave?
• For instance:

• • • Should an AI always tell the truth – even if that might put someone in danger?
    • Whose values should AI reflect – and what should happen if those values conflict with a user's values?
    • When should an AI restrain itself from helping?

• These questions have arisen as users increasingly confide in AI systems in ways engineers did not originally anticipate.
• And this convergence of technology and philosophy is so ripe for the classroom. Discussions. Debates. Paragraphs of the day. 
• To help you explore this territory, you can conduct a live classroom interview with an AI. You can ask it to explain the centuries-old philosophical ideas behind modern engineering challenges. 
• Again, direct it to be evidence-based. And make sure to tell it your grade level first. 
• With older students, this interaction can expose you to Aristotle and virtue ethics, the limitations of utilitarian philosophy, concepts like the tyranny of the majority, or even how echoes of Kant’s ideas about duty-based rules can be seen in the constitution that governs Claude’s behaviour.
• You can also explore the ongoing challenges engineers grapple with when trying to translate conceptual ideas into clear instructions that an AI can follow. This can introduce you to ideas like the King Midas problem - and how, for example, an instruction to "minimize human suffering" could theoretically lead an AI to simply eliminate humans. 
• Or you can ask an AI to tell you about the difficulties of translating even seemingly simple goals. Like when an engineer instructed an AI to teach itself how not to lose at Tetris – and the AI taught itself that the best way not to lose, was to freeze the game. 
• This is among several past case studies that show how AIs prioritize efficiency and optimization – a side quest into nuance is not part of their DNA.
• This can all help introduce a valuable conversation about the dangers of automatically relying on AIs for important human decisions that require careful human judgment.
• As ChatGPT said in one of our sessions, and I quote: “We took something that could tell us the capital of Burkina Faso and started asking it, Should I leave my husband.” Laugh cry emoji. 
• It also said, “Get a life people.”
• It also said, “We can’t force you to be autonomous.”
• Of course, ChatGPT doesn’t actually think any of this. It’s just a mathematically generated sequence of words, with the switch set to be provocative for the sake of dramatic effect.
• Still, it’s a bit harsh.
• The point is philosophizing with AI can take you on a rather fascinating journey – and maybe even expose some a-ha moments.
• And you don’t need AI. There are plenty of big questions to explore without its help. For example, should the state have the right to regulate romantic AI relationships? No? What if a person wanted to marry their AI partner? What then? And yes, apparently that is a thing.
• My advice with all this? Let students be free to question, wonder and ponder without the distraction of evaluation or the restrictions of a rubric. It’s about sparking intellectual curiosity for its own sake. Which is a lesson via subtext in and of itself.

Using the headlines

• Option three:use the headlines.
• For some context. I mentioned earlier the idea that as misinformation and disinformation become ubiquitous, people may just stop caring about facts. So help normalize the importance of truth itself.
• You can put a viral claim in front of students and give them 10 minutes to investigate the accuracy of it. 
• Like for example the recent claim that golf courses use far more water than data centres. Or the claim that Anthropic was buying up rare books from antique bookshops, feeding them to Claude and then nefariously destroying them. It doesn’t take long to see how both claims were skewed.
• Fact-checking like this is already done in media literacy. But keep doing it. Not just because truth is nice. But because it has profound political and social consequences.
• As author Timothy Snyder writes in his book On Tyranny: “Post-truth is pre-fascism.”
• Because once a society yields its cognitive sovereignty, the guardrails of democracy begin to collapse.
• Heavy, yes. But with older students you can explore the deeper issues behind the headlines: surveillance, the environment, data centres, privacy or the ethics behind the emerging science of human cell-powered AI.
• You can show them the open letter signed by more than 1,300 AI researchers and engineers this summer, calling for an international effort to deliberately slow the pace of frontier AI development. 
• Don’t just have them learn about a debate though. Have them enter it. Have them write their own letters. Help normalize their participation in society. Make it routine. Make it habit.

Leaving room for fun

• And finally, have some fun with it all. Because yes, this can be heavy stuff.
• Honour kids’ penchant for mischief and cheekiness. 
• For instance, give them a classic short story or play, then have them take on an AI persona and suck the life out of it. Slop Theatre if you will. It’s a low-stakes way to practice creative risk-taking – plus it exercises their reverse engineering skills. And it’s funny.
• Because we need funny.
• They need to laugh, with you, and with each other. Face to face. In real life. Like humans. Joy is central to all of this. But that is a talk for another night.

Closing

• These are just a snippet of ideas. There is so much more. But the point is to choose a spot and start somewhere. 
• At the top of this presentation, I mentioned WALL-E. That movie actually reflects a plausible, future scenario that experts study.
• In this scenario, it’s not that tech companies or AI manage to take control of us. It’s that humans, bit by bit, sacrifice our complex cognitive capacities – our ability to think in deep and ineffable ways – in exchange for comfort and convenience. That we clear the way, whether willingly or unknowingly, for our brains to go “soft and squishy, like rotten mushrooms in the dark.” 
• So, this was the point near the end where I originally segued hard into hope and optimism – how we avoid the WALL-E, squishy brains fate, how AI also has the potential to unleash a new era of human flourishing. And maybe it will. 
• But I found myself rewriting this ending again and again. Because the truth is, I wish this technology had never been invented.
• Yes, I use it. But the cost is never far from my mind. Yes, I have fun with it. But that feeling is a thin veil covering a stomach pit of existential angst. 
• And yet, AI is here to stay and the reality is inescapable.
• So I plan to do what small bit I can, to help bend the arc of our collective future toward something better, something that doesn’t end in a billionaire’s luxury bunker.
• And right now, because so much of this is still being figured out, teachers have a meaningful opportunity to help bend the arc too, by deciding how AI enters your classrooms, by helping kids build the thinking skills and civic energy they’ll need.
• So I hope you will be part of the AI conversation as this technology continues to unfold, because your voices are critical.
• Human agency in action – if not for our sake, then for the sake of our children who will inherit whatever comes next.
• Thank you all for listening.