Studies suggest that students’ unregulated use of AI carries significant risks to their cognitive development and can worsen equity gaps (Lodge & Loble, 2026). The convenience of AI can entice students to outsource the very effort they need to exert in order to learn – a process known as “cognitive offloading.” Because our students are still developing their skills and their ability to learn, instructors need to carefully weigh when and how they use AI with students.
The strategies below embody what research tells us about how people learn and help structure student use of AI. The underlying design principle behind these strategies is the same: students do the work first, generating their own explanations and answers, and use AI to help reflect, evaluate, and deepen their understanding.
This strategy has students use AI to reflect on their own beliefs and logic. Instead of providing answers, AI responds to students’ input with questions that help deepen their understanding.
Student activity:
- Identify a course concept you are struggling with.
- Prompt AI: I’m a college student trying to understand [insert concept] in [insert course title]. Act as a Socratic partner to help me improve my understanding. Don’t give me answers. Don’t correct me or explain anything. Ask questions that guide my thinking step by step. After each of my answers:
- Ask a follow-up question that builds on my reasoning
- Point out gaps or inconsistencies
- Engage in at least 3-5 rounds of questioning.
- Submit a reflection on the following questions:
- Which of the AI’s questions challenged you most?
- What about this concept remains confusing or unclear?
*Adapted from Monzon, N., & Hays, F. A. (2025). Leveraging generative artificial intelligence to improve motivation and retrieval in higher education learners. JMIR Medical Education, 11, e59210. https://doi.org/10.2196/5921
In this strategy, students assume the role of instructor and “teach” AI a concept. By explaining the concept in their own words, students can test and refine their understanding.
Student activity:
- Identify a course concept you want to “level-up.”
- Explain the concept to AI with the following prompt: Act as a cognitive mirror for [insert concept] in a college-level course on [insert course title]. I will explain the concept to you. Ask me probing questions one at a time and give me an opportunity to respond. Do not give me answers. Here is my explanation: [insert explanation]
- Respond to 3-5 rounds of questions posed by the AI.
- Review the chat transcript with AI and use it to create a better explanation of the concept.
- Submit your revised explanation of the concept, a copy of your chat transcript, and a brief reflection on how the exchange with AI helped you “level-up” your thinking about this concept.
*Adapted from Tomisu, H., Uneda, J., and Yamanaka, T. (2025). “The Cognitive Mirror: A framework for AI-powered metacognition and self-regulated learning.” Frontiers in Education, Vol. 10. doi: 10.3389/feduc.2025.1697554.
In this strategy, students use AI as a means of building information literacy skills. By asking students to evaluate and verify AI output, the strategy helps students develop critical thinking skills and healthy AI habits.
Student activity:
- Complete the UW Libraries Undergraduate Researcher tutorial on “Information Sources & Databases.”
- Prompt AI: Generate a bibliography of 10 articles on [insert topic] relevant to a college-level course on [insert course title]. The articles must be [insert any criteria provided by the instructor (e.g., peer-reviewed, from the last 10 years, etc.)]
- Evaluate each source in the bibliography generated by AI using the following questions:
- Who is the author?
- Were you able to find the article in any UW Libraries databases? Did the key bibliographic information for each source (e.g., author name, journal name, publication date, etc.) match the bibliography information provided by AI?
- Is this article relevant to your topic?
- Does the article meet the assignment criteria (e.g., peer-reviewed)?
- Submit a reflection on the following questions:
- Did the bibliography generated by AI provide a diverse range of perspectives and sources? What do you think might be missing from this bibliography (e.g., different themes, certain perspectives, different types of studies, etc.)?
- How did this experience shape your thinking about using AI to generate a bibliography?
*Adapted from Grace, S. L. (2025). Finding equilibrium: An integrative approach to balancing human and artificial intelligence in legal research. Legal Reference Services Quarterly, 44(1), 1–32. https://doi.org/10.1080/0270319X.2025.2534229
Please note: The prompts included in the strategies above will likely need to be adapted to your teaching context. In addition, because of the nature of generative AI, there will be a great deal of variability in the results generated by different LLMs. Finally, remember to always review and revise any work you do with AI.