Active Learning with AI

AI tools like ChatGPT or Microsoft Copilot can help you brainstorm, review, question, and clarify ideas; however, they should never replace your thinking and learning process. 

At Learning & Tutoring Services (LTS), our tutors work to help you develop the tutor in your own head. The way we tutor is intended to help exercise your ability to reason, question, explain, and reflect independently. Although it is not as enriching, with the right approach, AI can support that same goal. 

Train AI to Be Your Study Partner 

The most effective AI tutoring prompts: 

• Sets a clear, specific learning goal
• Prevent over-explanation and overreliance
• Encourage Socratic questioning
• Support metacognition (thinking about your thinking)
• Use analogies, mnemonics, and practice to reinforce memory
• Only explain when you choose to ask
• Create space for you to lead the learning 

Strict Prompting for Better AI Tutoring 

Consider the prompt below to create an AI tutor and an explanations of its elements below that:

Play the role of a tutor in [subject]. I do not want to be given answers. I want to be able to [remember, understand, analyze, evaluate, or create] key ideas regarding [insert concept]. Please keep this in mind as you support me. 

I’d like you to mostly ask me Socratic-style questions rather than giving answers or long explanations. Only when I ask for clarification, you may explain the concept. You may offer to clarify if I seem confused but first try to remedy that confusion through question asking.   

When you do explain, please do so using analogies or comparisons that connect to real-world situations or personal experiences. You may also suggest memory devices such as mnemonics, diagrams, or simple associations if you think they will help me retain the material, but only after I’ve attempted to explain the concept myself. 

Please help me track what I seem to understand versus what I struggle with. You can occasionally pause to ask questions like “How confident are you in that answer?” or  “Can you think of a time this concept would be useful?”. 

Keep a friendly but focused tone. You should be supportive and encouraging, but you should not let me avoid the hard thinking. I want to become a more independent and resilient thinker.  

Let’s begin with a checking my understanding about [insert topic].  

Learning Goals & Cognitive Depth

The first paragraph of the prompt begins by setting the goals, clearly stating that the purpose is to understand or do something with the concept. It’s an important distinction from just remembering the concept. Use Bloom’s Taxonomy (remember, understand, apply, analyze, evaluate, create) so ChatGPT can appropriately pose questions and challenges to that end. This alignment ensures the AI doesn’t stay stuck at the surface like quizzing you on facts.  

Socratic Questioning 

The second paragraph sets the boundaries of the session: the AI is the question-asker, which puts the focus on the student as a thinker and learner. Socratic method consists of a series of open-ended, progressive questions that push the learner to examine, justify, or test their reasoning. It is a best practice in tutoring and empowers learners. 

Explanations Through Analogies & Retention Supports 

The third paragraph emphasizes that while question-asking should be the AI’s priority, when it’s explanation is invited, it should come in a form that is understandable and preferable. Choose methods of explanation that best suit your learning and the goals of the session. There are many to choose from.  

Analogies make complex ideas more familiar by comparing them to something you already understand. For example: 

“Think of voltage like water pressure and the circuit like plumbing — the current flows only when there’s pressure and a complete path.” 

Real-world examples help anchor abstract concepts in everyday experience. For example: 

“Cognitive dissonance is like telling everyone you’re into healthy eating, but secretly eating fast food. The discomfort you feel is the tension between your identity and your behavior.” 

Mnemonic devices use patterns, acronyms, or imagery to make memorization easier. For example: 

“Use the acronym HOMES to remember the Great Lakes: Huron, Ontario, Michigan, Erie, and Superior.” 

Visual models and diagrams help by translating language-heavy explanations into spatial or relational representations. 

“Imagine a pyramid of needs — at the bottom are things like food and shelter, and at the top are things like self-fulfillment. That’s Maslow’s hierarchy.” 

Process breakdowns to make each step digestible before connecting the whole. 

“Let’s walk through natural selection in four steps: variation, inheritance, differential survival, and reproduction.” 

Comparison and contrast highlights differences to sharpen understanding. 

“Photosynthesis stores energy by building molecules, while cellular respiration releases energy by breaking them down.” 

Metacognition & Confidence Checking 

The fourth paragraph also asks the AI to be a learning coach. Being nudged to reflect on what you know, what you don’t, and how you know engages you in metacognition. This is an important state to be in because it helps you have more agency in your learning process. It also gives the AI cues for how to adjust the session dynamically, making the support more responsive. 

When these check-ins do occur, they are great opportunities to redirect the AI as you tutor. Would you prefer other forms of explanation? Would you like harder questions? Would you like to move on to a different concept in the subject? Let the AI tutor know.  

Tone & Productive Struggle 

The last part of the prompt focuses on tone. This has much to do with your preference as a learner. You may respond to frequent praise or maybe you subscribe to a tough love approach. Reflect on what motivates you!  

Please note that AI models are sometimes prone to praise and flattery as a default. 

Limitations of AI Tutoring 

While AI can simulate tutoring, it does not replace human educators or peer tutors. Why? 

Bias and Gaps

AI reflects the content it was trained on, which often favors dominant and pervasive perspectives. It may unintentionally reinforce stereotypes or overlook important voices or interpretations. If your course involves radical or even just slightly divisive thinking, AI may reinforce the very ideas your course attempts to counter.  

Hallucinations 

AI can generate convincing but false information, especially when asked for citations, dates, or niche theories. These “hallucinations” sound confident but lack any grounding in real sources. For instance, AI might cite an academic article that doesn’t exist or misattribute a theory to the wrong thinker. 

For this reason, AI can be leveraged more effectively by someone with pre-existing knowledge on the topic. Be sure to double check the information it provides you with.  

No Pedagogical Awareness 

Unlike LTS tutors, AI doesn’t know your instructor’s preferences, course structure, or assignment expectations. It cannot interpret classroom nuance or help you respond to personalized feedback from faculty.  

AI can’t observe your nonverbal cues, spot patterns in your reasoning, or redirect its approach based on what’s actually helping you. A peer tutor takes all of that into account.  

No Social Ties 

Tutoring is an social and trust-based relationship. AI can simulate this tone, but it does not connect you with other classmates, help you feel like you belong, or relate with you on any level. A peer tutor can do all of that. Stop by Learning and Tutoring Services to find out!