Teaching to Learn
Jody Ondich
AI as a Tutor
Using AI as a tutor can offer learners personalized, on-demand support that can adapt to individual learning needs and paces. By providing explanations, feedback, and practice opportunities, AI can supplement traditional instruction and reinforce student understanding. This approach has the potential to increase access to academic support while promoting self-directed learning.
Example: Feynman Technique
Goal
The Feynman Technique is a learning strategy that involves teaching a concept in simple, clear language as if explaining it to someone unfamiliar with the topic. By breaking down complex ideas into plain terms, learners identify gaps in their understanding and reinforce their knowledge through active reflection and simplification.
Here’s a breakdown of the technique:
- Choose a Concept: Select the topic or concept you want to learn and understand deeply.
- Explain in Simple Terms: Write down everything you know about the concept, as if you were teaching it to someone with no prior knowledge.
- Identify Knowledge Gaps: As you explain, you’ll likely identify areas where your understanding is weak or incomplete.
- Fill the Gaps: Go back to your source material (textbooks, lectures, etc.) and re-learn the areas where you had gaps in your understanding.
- Simplify and Refine: Once you’ve filled the gaps, re-explain the concept in even simpler terms, using analogies or examples to make it easier to grasp.
How It Was Done
The exercise is designed so that an instructor uses AI to tutor a student on a specific topic, with the goal of deepening the student’s understanding. The student starts with foundational knowledge, after which the AI poses targeted questions to stimulate critical thinking. As the student responds, the AI provides detailed explanations and constructive feedback, fostering an interactive and personalized learning experience.
For this example, I started with this prompt:
I would like to study for a test on Judaism, using World Religion: the Spirit Searching/Judaism as a source. I would like you to work as a tutor, generating appropriate questions for me to respond to, and giving me additional detail and feedback when I have answered these.
Copilot responded by generating these questions for me to answer.
- What is the primary Scripture for Judaism, and what are its main parts?
- Who is considered the Patriarch in Judaism and why is he important?
- What is the covenant that God establishes with Abraham and why is it important?
- What is the significance of Exodus in Jewish history?
- How did the 10 Commandments impact the identity of the Israelites?
These are pretty good questions. So I answered them with 2-3 sentences each. When I was correct, it affirmed that I was correct, but also filled in with more detail added to my answers. I intentionally answered #4 wrong, as my students often mix up the Exodus and the Exile in the Hebrew story, and Copilot got it right, and was thoughtful in explaining the differences between the two, using dates, geography, and other detail to clarify.
Results
This process is a fairly simple one, and works especially well if you provide the original learning materials for Copilot to use. In this case, giving it a chapter of a text allowed the AI to create solid questions for the student to answer. One could use library materials, publisher materials, or OER materials to submit to AI in order to get questions for student reflection and tutoring. I might have added a few more questions to these, but none of the questions provided were inappropriate, and the added information after I answered them was accurate and helpful.
Potential issues
- Students might be tempted to use AI to generate answers rather than engage with the material, which could limit deeper learning. Sustained motivation and academic integrity remain essential for this approach to be effective.
- Generating effective questions with AI can be more challenging without access to digital materials. This approach works best when students have their own notes or the original content available in a digital format to input into the AI.
Best Tools for the Job
- In this fairly simple situation, Copilot worked very well. In addition, Copilot protects student privacy and the copyright of materials used for generating study questions.
About the author
name: Jody Ondich
institution: Lake Superior College
Jody teaches philosophy at Lake Superior College in Duluth, and is part of a faculty development team there. Her focus areas are OER development, digital accessibility, AI use, and online course review programs. She has written 3 OER texts, housed at the University of Minnesota Open Textbook network.
