The Promise of AI Tutors
Cultivating success through the right learning culture...
👋 Hi there, welcome to The Learning Dispatch! We explore creative formats from the world of learning. Expect a note from us twice a month. Created by TeamLogiQ.
If you’re here for the first time, welcome! Feel free to catch up with our previous editions:
🔗 Learning Through Gamification
Now, on to this week’s format!
At TeamLogiQ, we're collaborating with a client on a new, exciting experiment: an AI Tutor. We're training a tool that uses AI to provide learners with an interactive, engaging, and personalized learning experience. Working on this project, which is designed for highly self-driven learners, has made me think if AI will eventually replace human tutors and teachers…
Research shows that pushing students to generate answers and think through problems is a major factor in the benefits of tutoring. Those are things AI does well, which is why that’s the learning format we’re exploring today...
📰 What’s the format?
AI Tutors.
How you feel about what AI can do kind of depends on what you think about human tutors. If you don't think human tutors do much, it's hard to see how AI could do any better. But if you think human tutors are helpful, then even AI that's not as good could still be very powerful, even if it's only half as effective.
Henrik Karlsson, whom we’ve referenced before in The Learning Dispatch, wrote a post that looks at the childhoods of exceptional people. ~70% of the people he researched had been homeschooled or tutored by parents, relatives, and governesses.
Mozart’s father was a devoted tutor to his children, with a deep love for music.
Marie Curie’s father built a laboratory in their apartment so they could study chemistry.
John von Neumann’s father would get so excited about their discussions that if they were, say, talking about machine weaving, he would set out to find a Jacquard automatic loom they could study.
Tutoring is effective because learners get more opportunities to restate ideas in their own words, explain, think out loud, answer questions, and elaborate on responses…1 They benefit by paying close attention to a skill or topic, actively working through problems, and getting immediate feedback as they make progress.
🎛️ What are the features?
Human tutors do some things exceptionally well. And in this article, Ethan and Lilach Mollick explain that a good AI Tutor do exactly the same. It:
Meets students where they are by asking them about their learning level, the topic they would like to learn about, and what they already know about the topic.
Provides explanations, examples, and analogies to clarify concepts, since students learn best through tailored explanations and diverse examples.
Guides students in an open-ended way and asks leading questions so that students have to explain their thinking.
Adjusts to student performance. If a student is struggling, it can offer additional support.
Prompts students to construct knowledge. It can end its responses with questions so that students keep generating ideas.
Asks students to explain the concept in their own words because being able to explain something yourself is a mark of understanding.
💡 Why is it effective for learning?
Tutoring is inherently interactive and involves a number of learning strategies including:
Questioning (by both the tutor and the student)
Personalized explanations and feedback (the tutor can correct misunderstandings in real-time and provide targeted advice based on the student's unique needs)
Real-time adjustment (based on the student's responses and progress, a tutor may adjust the pace or difficulty level, making the learning process dynamic and responsive).2
Good tutors ask questions and scaffold students’ thinking rather than just giving them answers or explanations. They create interactive dialogue, prompting students to respond actively; they check for understanding and guide students to construct their own knowledge; and they customize explanations based on what they know about students and their experiences.3
*A clear worry with AI tutoring is relying on a tool that might generate convincing but wrong answers.It’s key that learners are aware of these limits.
🥁 Exploring an example
I recently read a great review of traditional Khan Academy and its AI-powered tutor, Khanmigo. It compares them on three criteria: efficiency, effectiveness, and enjoyability. Also, it was written by Daphne Goldstein, an 8th grader. Here’s the bottom line in regards with Khanmigo:
Efficiency: You can learn a lot more in a lot less time.
Effectiveness: You can learn a lot more in depth.
Enjoyability: This criteria had mixed opinions. Some would enjoy the engaging, thought-provoking parts of an interactive AI, while others would rather just do work independently.
Sal Khan, the founder of Khan Academy, said he wanted to create a system to help guide students, rather than simply hand them answers. This is why Khanmigo is engineered to use the Socratic method: it often asks students to explain their thinking as a way of nudging them to solve their own questions. In other words, it’s trying to get how students think and why they think that way. This is different from usual school stuff that mostly cares about getting the right answer, instead of making sure you really get the ideas behind it…
💭 What’s Our Take
So… Will AI replace human tutors and teachers?
Some say yes (like Luis Vohn Ahn from Duolingo, whom we talked about in our previous edition, or the guys behind Synthesis School, who developed a tutor that’s custom-built with decades of DARPA research).
Some say no (like Dan Meyer, whose Substack I highly recommend, or Audrey Watters, who emphasizes the importance of human interaction and the social aspects of learning that cannot be replicated by machines).
I really liked Sean Geraghty and Mike Goldstein’s perspective on this. They claim that an AI Tutor is a tool for two very different situations.
Motivated learners will increasingly substitute AI for human tutors and teachers; and
Unmotivated learners will rarely (of their own accord) do this.
So they basically formulated a new questions altogether (and a better one, I think), which is: What if AI has profoundly different effects on motivated and unmotivated learners? Which brings me to another question (if you’re still here, thanks for joining me in this rabbit-hole!): What keeps learners motivated?
My take is that a big part of it is creating the right cultures of learning. Creating a culture of learning goes far beyond tools, frameworks (or prompts!) A culture lives in values and beliefs: focusing learners on the learnings vs the work, teaching for understanding vs knowledge, encouraging deep vs surface learning strategies, promoting independence vs. dependence…
Even the way we use language directs learners to specific cognitive acts! Having a language to identify thinking processes is a requirement for us to call them into play… (I might write a long(er) post on this, if it sounds interesting to you).
Cultures and tutoring systems are complements. The more powerful AI tutors become, the more valuable cultures that support learning will be. ~ Henrik Karlsson
As we continue navigating the discussions around AI's role in education, it's crucial to remember the ultimate goal: cultivating a culture that empowers all learners…
🏷 Summary
The conversation surrounding AI tutors shouldn't focus on replacing human educators; instead, it should emphasize what learning cultures do we need to create to keep learners motivated. The goal of integrating AI into education goes beyond mere innovation—it's about fostering an engaging learning environment that captivates students.
At its core, the true measure of success for AI in education is its ability to ignite a lifelong passion for learning. This serves as a reminder that education's essence is deeply rooted in nurturing growth and empowering learners to reach their full potential.
📚 Further Readings
Assigning AI: Seven Approaches for Students, with Prompts (by Ethan R. Mollick and Lilach Mollick).
Here’s the prompt the paper refers to help any learner study any topic.
👉 Coming Up Next
We value your feedback (suggestions, critiques, positive reinforcement, constructive ideas…) as well as your tips or suggestions for future editions. We’d love to hear about you in the comments.
→ Or just click the heart symbol, that always makes us smile 🙂
Chi, M. T., Siler, S. A., Jeong, H., Yamauchi, T., & Hausmann, R. G. (2001). Learning from human tutoring. Cognitive science, 25(4), 471-533.
Chi, M. T., Roy, M., & Hausmann, R. G. (2008). Observing tutorial dialogues collaboratively: Insights about human tutoring effectiveness from vicarious learning. Cognitive science, 32(2), 301-341.