Teaching, Learning, and our Shared Responsibilities with AI
In May 2026, Pope Leo XIV published his first encyclical — a 42,000-word letter on artificial intelligence titled Magnifica Humanitas. That same month, I started preparing for one of Notre Dame’s summer engineering programs in London: six weeks, three credits, a new course on wireless technology and spectrum policy I’ll be teaching for the first time. And AI tools are reshaping everything about how students read, write, and think.
The encyclical opens with a choice: build a new Tower of Babel with AI — homogenizing, controlling, reducing persons to data — or build something else. I kept coming back to that image as I envisioned the class and designed my syllabus. Because the choice Pope Leo XIV describes wasn’t abstract. It was staring at me from my course outline.
So here’s what I’ve been working through about teaching and learning in the AI era.
Two jobs, one classroom
My job is to teach. A student’s job is to learn. Each of us gets out of it what we put into it.
If I hand my teaching off to AI — generate the slides, generate the discussion questions, generate the feedback — I shortchange my students and myself. The preparation is part of the teaching. The thinking I do before class is what makes me useful when I’m in the room. If I skip it, I’m not really there. On the other hand, perhaps I can prepare even better with strategic prompting and careful revisions, accelerated by AI.
The same goes for students. If students hand their learning off to AI — generate the summary, generate the response, generate the paper — they may produce something that looks like learning, but they haven’t done the work that produces understanding. They’ve outsourced the thing they are there to do. On the other hand, perhaps AI can help students develop more ownership for their learning, and enable them additional review and exploration support at a scale and level of personalization that one instructor like me could never sustain.
So if my students and I use AI to go deeper, to prepare better, to explore further, to spend our limited time together on what actually requires another human being in the room, I believe we can get more out of the experience than we ever could before. The tools are useful. The question is whether we’re using them to skip the work or to amplify it.
Content is becoming a commodity
For a long time, part of what a professor brought to a course was organized access to knowledge. I had read the textbooks and papers. I had synthesized the field and developed my own point of view. I knew what to assign, what to skip, and how to explain the hard parts. That is real value, and it takes years to accumulate.
AI has changed the economics of that dramatically. A first draft, a literature summary, a plain-language explanation of a technical concept — these are now available on demand, to anyone, essentially for free, in seconds. Raw content itself is becoming a commodity. On the other hand, quality content enriched from human review and interactions will always require human time scales.
I think this is mostly a good thing, even though it’s disorienting. It means that what being a professor is for has to shift. Not a gatekeeper of knowledge, but a guide through it. Not a source of all the answers, but a sharpener of questions. The commodity content can get a student to the threshold of a hard idea. What happens at the threshold is still the professor’s job, and it’s the part that matters most.
The hallucination problem
Here is the complication: AI makes things up. More than occasionally, and not obviously. It hallucinates, generating text that is confident, fluent, and wrong. A fabricated citation, a misattributed quote, a plausible-sounding technical claim that doesn’t hold up. If students don’t already know enough to catch it, they won’t.
The traditional model of education placed the instructor as the primary authority on truth. That model was already under strain because the pace of research in most fields means no single person can be on top of everything. With AI, the model breaks. I cannot read and fact-check every AI-generated passage every student in my course produces. It’s not possible, and pretending otherwise doesn’t help anyone.
What replaces it is shared responsibility.
Part of my job is to teach students how to think for themselves and catch errors. To know enough about a subject to recognize when something seems off, to verify claims against primary sources, to hold AI output to the same standard they’d hold a newspaper or journal article. Treat AI-generated output as a useful starting point, not the final word. My student’s job is to bring active skepticism rather than passive trust. Not to assume the AI is right because it sounds confident, but to ask whether what it generates holds up.
This is a new mindset and requires new skills for both of us. It requires intellectual honesty about the limits of what any one person, or any one tool, can know. In some ways, it’s more honest than the old model, which perhaps overstated the instructor’s authority and understated the student’s responsibility.
What doesn’t become a commodity
None of this applies to the learning experience itself.
The discussion where someone says something unexpected and the whole room shifts. The moment a student asks a question that makes me think differently about something I’ve been teaching for years. The conversation that starts after class and continues over the following week. The gradual realization — which takes time and can’t be rushed — that a concept students thought they understood is actually much stranger and more interesting than they thought.
AI can accelerate the work that leads to those moments. It can help a student get through the background reading faster, arrive at the discussion better prepared, explore a question further before class. All of that is real preparation. But the moment itself — the encounter between a mind working hard on something and another mind that has worked hard on it longer — that doesn’t get generated. It has to happen.
The lasting value of education is the experience, the intellectual maturation, and the relationships formed in the process. The content students cover will be superseded. The skills they develop for a specific tool will become obsolete. What stays is the habit of asking big questions, the confidence that comes from thinking hard, and the people they do it with.
What this means in practice
A few things I’m holding onto as I design my London course, and would offer to students and other instructors:
Design around the experience, not the content delivery. If a lecture can be replaced by an AI summary, it probably should be, which frees up class time for the discussions, demonstrations, and exchanges that can’t be replaced.
Use AI to prepare, not to skip preparation. The goal is to arrive more ready, not to arrive having done less thinking. There’s a difference between using AI to research a question before class and using it to avoid engaging with the question at all.
Treat hallucination-detection as a skill worth developing. Know the sources, and learn which ones can be trusted. Verify surprising claims. Build the habit of checking, not because AI is completely unreliable but because critical evaluation of any source, AI or otherwise, is what it means to actually know something.
Be honest about what you don’t know. I say this to myself as much as to students. I’m teaching a course I’ve never taught in a city where I’ve never taught, on a topic that is moving faster than any syllabus can track. I won’t know everything. Modeling what it looks like to not know something, and to go find out, is part of teaching.
The frame that stuck with me
Magnifica Humanitas makes a simple claim: technology is never neutral. It takes on the values of those who use it. The same AI tool can be used to flatten education into a content transaction or to deepen the parts of it that only humans can give each other.
The phrase from Pope Leo XIV to which I keep returning is: “Our pressing duty is to remain profoundly human.” That’s not a warning against AI. It clarifies what the job actually is, for both the instructor and the student.
My London course starts July 1. I’ve drafted the content, designed the AI explorations, structured the discussions. That content will be revised with acceleration from AI many times. What I hope my students carry out of that room — from the conversations in Fischer Hall, on the walks between field sites, in the thinking they do after class — is something the encyclical is reaching for. Not the commodity. The encounter.
AI Notice: The development of this post was accelerated with the help of Anthropic Claude and Google Gemini.