Using AI tutor in philosophy class leads to deeply human conversation

Lisa Walker


In Professor Mark Collier’s philosophy class, innovation comes in the form of an artificial intelligence tool, or bot, that acts as a digital tutor to help students further engage in the class and gain a deeper understanding of the material covered. With a background in cognitive science and philosophy, Collier recalls when artificial intelligence emerged in the 1990s. His work then centered around the philosophical consequences of neural networks. Now he is using AI to help his students get more out of the class material.

“I wanted to figure out a way that AI could promote the learning outcomes of my course,” says Collier.

As part of his 2024-25 University of Minnesota Emerging Technologies Faculty Fellowship, Collier developed an AI tutor. Collier used the custom GPT technology that OpenAI had recently introduced. The tutor is trained on readings from his Ethics and AI course and given particular instructions about how to interact with students in a conversation.

The goal of the fellowship is to foster a learning community that experiments with generative AI  in their teaching and to promote the effective use and best practices of AI. Fellowship faculty are tasked with identifying an area of AI to explore and implement within a university course.

Side view of Mark Collier from waist up wearing black jacket, talking to students
Professor Mark Collier

“I felt that [creating this AI tutor] might be a good use case of generative AI tools.”

Collier was cautious that his students use AI in a way that strengthens their independent and critical thinking skills.

“The … tutor is Socratic in the sense that, like the philosopher Socrates, it only asks probing questions but never answers them.”

Collier chose to model it on Socrates who encouraged those he conversed with to think more deeply about their own views and the reasons why they believe what they do.

The AI tutor is instructed to figure out conversational strategies that will lead students to a deeper understanding of the text they are studying. Collier noted that students soon found that they had a better experience interacting with the tutor when they were more specific with what they wanted to ask it. He points out that this had the added benefit of teaching students how to formulate questions, which he says falls in line with the goals of a liberal arts education.

“[I]f you don’t ask a clear question, you aren’t going to get a useful answer,” says Collier.

So just how do students use the AI tool? Collier’s class met three times per week. On Mondays and Wednesdays, the class was part lecture, part discussion about a text the students had read. Between Wednesday and Friday, Collier had students choose a topic from the readings that they wanted to explore further. They used that topic to engage in a discussion with the AI tutor. Then they were to post on the class’s online Canvas forum a link to their original question, the conversation, and a reflection about whether their views changed after the conversation. On Fridays, each of the topics was discussed as a class.

Collier says the students really enjoyed the Friday discussions. “[T]hey always emerge with a much deeper perspective on the topic.”

2 students seated next to each other at a classroom table with laptops open, one talking

Student Ryan Hyers ‘26 says his experience with the AI tutor was positive. He found it a very useful and an exciting way to explore class concepts further. He says the back-and-forth discussions with the tutor enhanced his knowledge of the course material.

“[The bot] helps explain and interpret it in ways to better help my own understanding. There have been multiple occasions in which my discussions with the AI tutor have allowed me to better develop my own ideas by [the tutor] presenting counter arguments to me or other variables that I hadn't previously considered.”

Student Angela Albrecht found that the tutor helped her “think more thoroughly,” leading her to a more well-rounded conclusion on each topic.

“I like how the AI tutor helps me reach a conclusion by having me answer questions about my own questions,” says Albrecht. “Instead of just coming to a conclusion, I can understand how I got to that conclusion.”

She adds that for her, the tutor triggered a closer examination of solutions. “[I]t helps me … think through my conclusions so that I understand my own thoughts better.”

Collier says the students seemed to look forward to bringing their conversations to the attention of others and discussing them as a class. This leads him to believe that, while AI may very well play a larger role in the future of education, conversations with AI bots will not replace in-person instruction.

“[A]fter you have a conversation with a tutor bot, it makes you want to talk to your fellow—human—students even more.”

The class last fall was the first to use the tool and Collier says it is still a work in progress.

“I've continued to refine and adjust it based on how it performs with students.”