Key Takeaways

• In the Stanford study looking at the results of human tutoring, they found that students who had a tutor who utilized AI and would also be considered a "sub-par" tutor saw an almost 10 percentage points improvement in mastery of the subject being studied.
• The context and training of the AI is the important factor when looking into the studies on this matter. There is not always a gain when students use AI (alone or in conjunction with a human tutor) but when the model is trained to use pedagogy and methods that are proven to work (like spaced repetition, being mindful of cognitive load, scaffolding, etc) the students showed measurable improvement.
  • Prompting students to think and explain is FAR superior to giving them answers to digest.
• The human element is still important. Much like how the relationship between a therapist and patient is an important part of the therapeutic process, so too is the relationship between student and tutor. The student still needs the connection and empathy of a human tutor to navigate some parts of the process.
• The answer, like many things in life, is neither is better alone but do best when used together.

Atomic Idea Sparks

• Metacognition is an effective tool for learning - It isn't about how long you spend reading, how many times you re-write your notes, or whatever else you do. At the end of the day, one of the best things you can do is evaluate why you think the way you do. What led you to this understanding? What schemas are you pulling from? By teaching students how to think critically and go through a process we are teaching them to interact with the world, with the knowledge, so they are able to use it, not simply regurgitate it.

Critical Reflection

• I want to find the actual meta analyses that were done because I have questions about what caused the lower rated tutors to have such success with their students simply because they incorporated the use of AI. What is the game changing element there?

See Also