This summer, I had an opportunity to attend a physics conference at the University of Notre Dame in Indiana. It was a special experience. Notre Dame has one of the most spectacular university campuses in the United States, and simply walking around the campus offers many opportunities to explore, observe, and learn. During one of my walks, I visited the university’s art museum: Raclin Murphy Museum of Art. Perhaps because it was summer, the galleries were practically empty. I suddenly found myself surrounded by remarkable works of art, with plenty of time to look at them ,but without enough knowledge of religion, art history, or even literature to fully appreciate what I was seeing.
So since I was alone in the halls of this magnificent museum, I decided to try something different: I invited AI to become my personal museum guide. I photographed paintings that caught my attention and a
sked ChatGPT to help me understand them. Who were the people depicted in the painting? What story was the artist telling? Why was a particular object placed in the corner? What should I notice about the composition, gestures, clothing, or symbolism?
What surprised me was not simply that AI could provide information. It was how the interaction changed the way I looked at the paintings. Instead of glancing at a painting, reading the label, and moving on, I began to look more carefully. ChatGPT would point out a detail, and I would return to the painting to find it. That detail would lead to another question. Sometimes I challenged the interpretation. Sometimes I asked for clarification. Sometimes the answer reminded me of something I had learned years ago and almost forgotten. Many times I asked to translate specific terms into Russian or Hebrew, so I can make connections. In other words, AI did not replace my looking. In some ways, it made me look more carefully. It also allowed me to build bridges and make sense of what I was seeing based on my own personal experiences.
Yet as I walked through the nearly empty galleries, another question kept coming back to me:
Would I have learned more if I had been accompanied by a knowledgeable private guide?
Almost certainly, a knowledgeable and inspiring human guide would have offered something different. An expert standing next to me could notice where my attention was directed, recognize my confusion, respond to my facial expressions, tell stories based on years of experience, and perhaps challenge my interpretation in ways that AI could not. But there was another side to it. Would I have asked a human guide all the questions I asked ChatGPT? If I were in a group, would I feel comfortable asking personal questions and taking time from everybody else? Probably not.
Some of my questions were very basic. I repeated myself. I asked for explanations in different ways. I moved between art, history, mythology, religion, and language. I could follow a small detail simply because it interested me, without worrying that I was taking the conversation in the “wrong” direction. There was no embarrassment in admitting that I did not know something that perhaps I should have known.
And that realization brought me back to something I think about constantly as a STEM educator.
What if AI could play a similar role in learning science?
Imagine a student working through a physics problem. The student does not understand why the normal force is not always equal to (mg). Or perhaps she has obtained an answer but does not know whether it makes physical sense. Maybe she understands the mathematics but cannot visualize what is happening. Traditionally, she might raise her hand and wait for the teacher. She might ask a classmate. She might search online. Or, very often, she might simply move on without asking. Now she has another possibility. She can ask AI.
But this is where the analogy with my museum experience becomes important. The educational value was not that AI could tell me what the painting meant. The value came from the interaction between what I was observing, what I already knew, what AI suggested, and the new questions that emerged as a result.
The same distinction matters enormously in STEM education. If we use AI simply to provide students with answers, we may make learning faster while simultaneously making it shallower. A student who asks, “Solve this problem for me,” may receive a beautiful solution without doing much thinking. But what if we inspire students to use AI differently?
“What should I notice here?”
“Don’t solve this problem. Ask me a question that will help me figure out what I am missing.”
“Here is my reasoning. Where might I have made an incorrect assumption?”
“Give me a hint, but not the answer.”
“Can you show me another way to think about this?”
“Does my answer make physical sense?”
These are very different uses of AI.
Perhaps the most interesting educational role for AI is not to become an answer machine, but to become a thinking companion, one that is always available, endlessly patient, and willing to engage with questions that students may hesitate to ask another person.
Of course, my museum experience also reminded me of AI’s limitations. AI can be wrong. It can confidently interpret something incorrectly. It does not truly see my confusion in the way an experienced teacher or museum guide might. And it cannot replace the expertise, judgment, empathy, and human connection that an excellent educator brings to learning. AI also can hardly make somebody interested in exploring things deeper. This is where an inspiring Teacher would come in.
So perhaps the question should not be:
Can AI replace the teacher—or the museum guide?
That seems to me to be the wrong question.
A more interesting question is:
What kinds of learning become possible when a knowledgeable human teacher, a curious learner, and AI all become part of the same learning environment?
Walking alone through an almost empty art museum at Notre Dame, with my phone in my hand and an AI guide in my pocket, made that question unexpectedly real for me.
And I suspect that this is precisely the question we should be asking as we think about the future of STEM education.