AI represents both a dream and a warning for education. On one hand, it offers personalized feedback, instant support, and creative possibilities that can make learning more dynamic than ever. On the other, it tempts us toward speed over depth, efficiency over understanding, and convenience over curiosity. The real opportunity lies in the space between utopia and dystopia, where teachers and students use AI not to replace human thinking but to extend it. When we design with intention, AI becomes a catalyst for deeper learning rather than a shortcut around it.
If you enjoy this blog but you’d like to listen to it on the go, just click on the audio below or subscribe via iTunes/Apple Podcasts (ideal for iOS users) or Google Play and Stitcher (ideal for Android users).
One of my favorite social studies writing prompts requires two separate days to complete. On the first day, students design their own utopia. They begin with the question, “What is the ideal society? What values should it be built upon?” From there, they dream up this perfect situation. Students complete a concept map connecting every aspect of society, from transportation to education to energy to entertainment to city planning. Next, they describe a day in the life of this dream society using the five senses to create greater imagery.
The next day, they begin the second half of the writing prompt. This utopia has become a dystopia and the dream is now a nightmare. What happened? What is it like to live in this society? What were the hidden weaknesses of the systems, structures, and even values embedded in that initial idea? What happens when we take a good idea and move it to an illogical extreme?
While I designed this as a social studies thought experiment, I actually think we can apply the same process to our dream scenarios in education. In many ways, AI represents a kind of educational utopia: instant feedback, personalized instruction, limitless access to information. But just like those student-created worlds, our idealized version of AI has cracks beneath the surface that we need to explore.
At the same time, we tend to have a glut of dystopian stories about the dangers of AI. From The Terminator to The Matrix to Ex Machina, our culture treats artificial intelligence as an inevitable descent into a dystopian future. Machines rise up, humans lose control, and the technology we created to make life easier ends up erasing what makes us human.
In this vision, schools become untouched sanctuaries, sealed off from algorithms and automation. Yet even that version can slip toward its own dystopia: one where students are unprepared for the world they actually live in, where fear replaces curiosity, and where learning becomes nostalgic rather than transformative.
I mention this because I recently led an all day AI workshop with secondary teachers. They actually spent twenty minutes in teams creating their dream scenario (I didn’t call it a utopia) with AI based on their previous beliefs about rejecting or embracing AI. Afterward, they analyzed what might cause their dream to become a nightmare.
Together, we discussed how the real opportunity isn’t about creating a perfect system powered by AI. It’s about asking deeper questions: How might AI make learning more human? How might it help students think more deeply, not just work more efficiently? In other words, we were able to discuss through the lens of story, how we might avoid the two extremes of Lock It and Block It and Techno-Futurism.
The question isn’t whether AI will shape learning, but how deeply it will shape thinking. Between the dream of effortless perfection and the fear of total collapse lies a quieter, more human possibility. It’s the chance, albeit messy, to use AI as a tool for deeper learning. The challenge for educators is to design learning experiences where AI doesn’t replace curiosity or struggle but helps students go further into both.
We live in an era of rapid change. Technology is moving at a breakneck speed and our lives feel more crowded than ever. We can access rapid fire video on multiple apps and stream on demand shows on multiple platforms. It’s no surprise, then, that the number one thing I hear when I ask something how they are doing is the short answer, “busy.”
This hurried pace is a broader cultural trend, but teachers are feeling it more acutely than many other professions. Every shift in technology, every new expectation, and every “time-saving” innovation seems to add another layer of urgency instead of relief. The role of the teacher continues to expand beyond teaching and learning into zones once owned by specialists (just look at the pushback I got in this Facebook post for suggesting that secondary content experts might not need to be experts in early literacy), counselors, curriculum designers, etc. We keep piling more onto teachers’ plates. New standards for the already packed curriculum map. New unfunded policies and initiatives. Even the most dedicated educators feel stretched to the edge of what’s possible.
So, given this reality, many teachers have looked to AI as a time-saving device. That rubric that once took an hour to make? I can revise an AI-generated rubric in five minutes. That newsletter that used to take two hours to write? Let’s use AI to craft a new one that can then be differentiated for parental preferences (a short, “just the facts” version, a typical version, and a detailed version). That station rotation activity? Let’s design it complete with handouts, instructions, and leveled readers in half an hour.
Students are feeling this same pressure to move faster. Many of them spend hours completing homework while participating in regimented after school activities. Suddenly, AI arrives as a way to save time and take shortcuts. On the surface, that makes sense. AI tools offer instant feedback, personalized support, and quick answers that can genuinely help students learn more efficiently. And when the curriculum map moves quickly, this instant feedback and quick tutoring can make a huge difference.
But there is also a cost. When students rely on AI to do the hard thinking for them, they miss the slow struggle that builds understanding. Over time, that convenience can lead to cognitive atrophy, where curiosity fades and learning becomes something to complete rather than explore. In its worst iterations, students end up using AI to do all of the thinking and the work for them. This cognitive off-loading can erode thinking over time.
However, we can also take a different approach that focuses on deeper learning. Here, students engage in deeper problem-solving that leads to mastery. They focus on meaning rather than speed and output. They engage in deeper curiosity and critical thinking. In other words, they use AI in a way that gives them the depth advantage in life:
So as we think about integrating AI into our schools at the student level, we might want to start with the question, “How does this promote deeper learning?”
I am sitting here staring at the cursor blinking back at me in silent judgment at my slow, foggy brain and my clunky human hands. It is 2:56 pm. In three hours, I’ll be teaching my assessment course. But for now, I am sluggish. One part circadian rhythm. Another part being a desert boy in a cold and gray climate. But mostly? I stayed up until midnight watching an exciting baseball game.
You know who doesn’t get tired? Generative AI. It might gobble up electricity and use a fair share of water (though this is actually a really complicated environmental issue that goes beyond the binaries of good and bad). So, right now, I am tempted to pass the baton from my hands to a complex algorithm built on predictive analytics.
But I don’t.
Part of it is my love of writing. I love the process even when I don’t enjoy it. I find value in the process even when it moves slowly. I define myself as a writer and I am convinced that part of being a writer is continuing when you are tired. It’s slowing down and being deliberate with words. It’s the process of learning through writing rather than just writing after you’ve learned. I don’t want to lose any of that even if it saves time and energy.
On some level, though, we all need to negotiate what we gain and what we lose by using AI. For some people, the idea of AI jump cuts (video editing) takes out the slow, methodical eye that videographers need as they engage in film production. To others, it’s an unpleasant task that they can outsource to a machine. For some, AI-generated spreadsheet codes allow them to analyze data faster. For spreadsheet nerds (the self-proclaimed “freaks in the sheets”) that takes the joy out of making spreadsheets.
As a classroom teacher, you might love to make slideshows, design leveled readers, or write a newsletter. But you might just want to spend more time in small groups or engaging in professional learning. You might hate data analysis and love the idea of an AI tool that helps you get real-time data to pull students for intervention. Or that might just be something you love to do in order to get a sense of how each student is doing.
In other words, it is idiosyncratic. We all decide when and how we want to use AI tools. But we do need to recognize that we lose something human every time we use AI.
This is why it helps to take what Ethan Mollick calls the “cyborg” approach, where we treat AI as a form of co-agency. We use the AI but we modify what it creates based on our own voice, contextual understanding, and knowledge of our students.
I use the metaphor of ice cream. AI creates the vanilla but we add our own unique flavor.
Co-agency means we are not outsourcing our thinking but partnering with it. Mollick describes two main approaches to this relationship. The centaur approach treats AI like a teammate. The human and the machine divide tasks based on their strengths, alternating between intuition and computation.
The cyborg approach goes a step further. It is fully integrated, where human and machine thinking blend in real time. In this model, AI does not replace creativity or judgment. Instead, it extends it. We remain the authors, the composers, the teachers, but with an extra layer of insight or speed or perspective. The challenge is to stay grounded in our humanity while still exploring what this partnership can make possible.
Deeper learning with AI requires intentional friction in this co-agency process. It involves doing unpleasant but important tasks. It focuses less on efficiency and more on extended learning. It means designing moments where students wrestle with ideas instead of rushing to answers. When AI can instantly summarize, explain, or generate, the teacher’s role shifts from delivering information to creating conditions for curiosity and productive struggle.
Students might use AI to test their understanding, to compare interpretations, or to revise their reasoning, but they still have to think, question, and decide. In this way, AI becomes less of a shortcut and more of a mirror that reflects how well students can apply, adapt, and extend what they know. But this requires us to view AI as a thinking partner rather than an efficiency machine.
I really struggle with the idea of a AI as a thinking partner because machines don’t think. They generate scripts. They aren’t intelligent. They simply generate responses using predictive analytics. True, an LLM mimics the human brain but it is not sentient, intelligent, or affective. AI cannot understand anything. It can generate questions but it cannot become curious. It doesn’t wonder. So, in a sense, AI is a tool and not a thinking partner.
Yet there is still value in using the metaphor of a thinking partner, because it captures how AI can extend and challenge human reasoning even without genuine understanding. When we treat AI as a thinking partner, we are really using it as a reflective surface that pushes our thinking forward. The goal is not to share cognition with a machine but to use its responses as prompts for deeper human thought.
When students use AI as a thought partner, they can improve their metacognition. They can determine what they know, what they don’t know, and what they need to do next. Along the way, they can move through this metacognition cycle:
Moreover, AI can actually strengthen student agency when students use it in a slower, more intentional way. They can ask better questions, explore new ideas, and take more ownership of what and how they learn. Here they move from compliance and engagement and toward deeper empowerment.
A student might use AI to clarify a confusing concept, plan a project timeline, or get feedback before revising their work. The teacher’s role is to guide those choices, not to script them. When students learn to use AI as a tool for curiosity and reflection, they begin to see themselves as active creators rather than passive receivers of knowledge. In other words, students can use AI as a partner in curiosity or as a coach for reflection. So, let’s explore a few practical ways students might use AI for deeper learning tasks.
AI can become a powerful tool for inquiry when students use it to spark curiosity instead of just finding quick answers. It often starts by asking questions and follow-up questions. Students can move in a slower, more methodical way by using the FACTS Cycle for prompt engineering.
Here, students might ask questions that help clarify a misconception they have about a mathematical concept. They could interview a fictional historical figure, like a woman who lived in ancient Egypt. Students might even use an AI chatbot as the starting place for a Wonder Day project.
But that’s only one element of inquiry. If students want to develop deeper curiosity with AI, they can ask a chatbot to challenge their assumptions, offer counterarguments, or reveal perspectives they might not have considered. Here, it becomes a way to explore ideas more deeply.
At the same time, students also need to spend time analyzing the answers that generative AI produces. While LLMs continue to improve, they will still create hallucinations (inaccurate information that it produces based on bad “guessing”). Information will continue to be biased as well. This is part of why students need to explore the nature of AI rather than just focusing on how to use it.
When students use AI to wonder rather than to finish, it turns learning into an open conversation instead of a checklist.
One of my favorite uses of AI is as a study aid. Instead of having students re-read or underline the text (which fails to improve metacognition or lead to deeper retention of material), students focus on having the AI test them on the material. Here’s how the process works. Note that this still incorporates a human element built on personal agency:
- Make a t-chart of everything you know and don’t know about what you have been learning in a class (or in a unit of study)
- Provide AI with a prompt asking it to analyze your work. Be sure to give it a RAFT and to ask for it to list trends, in a chart, of what you know and don’t know
- Compare your t-chart to the chart from the AI and then identify 5 items you want to learn
- Ask for an informational text on the first item.
- Engage in a Q&A with the chatbot where you ask clarifying questions.
- Once you have a solid understanding, ask it to give you multiple choice questions that go from easy to hard. For each incorrect answer, ask it to give you an explanation and give you a chance to ask clarifying questions before continuing.
- At some point, have the AI move from multiple choice to open-ended questions where you type your answers.
- Review the next day and move to the second item of what you don’t know and continue the process
In the future, students will need to solve what are called “wicked problems.” These are the types of problems where the solutions are imperfect and often lead to new problems. These problems require human wisdom and deeper contextual understanding. They often involve empathy as well. In other words, they are the kinds of problems that require human skills in the age of smart machines.
Here’s where AI can help with the problem-solving process by running simulations that allow students to test ideas. In science, students can explore what happens when they adjust one variable in an ecosystem or change the rate of carbon emissions. They get to see the ripple effects and talk about why it happened, which builds real understanding instead of memorized facts.
In social studies, AI simulations can drop students into complex situations like peace negotiations or debates over economic policy. They can see how one decision affects another and how competing interests collide. This kind of work helps them think like historians and civic leaders rather than test-takers.
In math and engineering, students can use AI to design bridges, rockets, or roller coasters and test how small design changes affect outcomes. They learn through trial and error, reflection, and revision. Across subjects, AI simulations make problem solving more active and give students a safe place to explore, fail, and try again.
We are still in the early stages of these deeper simulations. However, this is one of the rare moments where we use AI in a way that moves past integration and into true transformation. See the continuum below.
As mentioned earlier, the student use of AI should actually improve rather than reduce metacognition. One way to do this is to use AI as a metacognitive coach. Students start by sharing an initial goal that they have and then ask AI for feedback on that goal using a SMART Goal approach. Generative AI tends to do really well in estimating how realistic a goal is, especially in terms of time frame. As humans, we get pulled into the planning fallacy, where we overestimate how much we will accomplish in a given time. But generative AI tends to do well in making accurate time estimations based on larger predictive trends. Similarly, students can use AI to help break down goals into smaller tasks and action plans. Then, over time, the AI takes on the role of a coach asking reflection questions that lead to deeper reflection.
Another option might be for students to ask for specific feedback. If we think about our receptivity to feedback, we tend to react emotionally to feedback based on whether we expected it, whether the feedback can lead to extrinsic punishment or rewards (impacting our grades, for example), or whether we asked for it.
When feedback is unexpected and unsolicited, most people get angry. When it is expected but unsolicited, we get anxious. When it is expected and solicited, we tend to embrace it more enthusiastically.
So, students can ask for specific feedback on their work or their ideas and they can actually ask specific questions that lead to action. They might create a set of fictional avatars that offer immediate feedback. They can then engage in a fictional conversation with the avatars where they ask for feedback that is both positive and critical.
Get the FREE eBook!
Subscribe to my newsletter and get the A Beginner’s Guide to Artificial Intelligence in the Education. You can also check out other articles, videos, and podcasts in my AI for Education Hub.
Fill out the form below to access the FREE eBook:
