Khanmigo doesn’t answer students’ questions directly, but starts with his own questions, such as asking if the student has any ideas on how to find the answer. Then he guides him to the solution, step by step, with hints and encouragement.
Despite Khan’s broad vision of “amazing” personal tutors for every student in the world, DiCerbo sees Khanmigo as a more narrow role as a teacher. When students are working independently on a skill or concept but are stuck or stuck in a cognitive rut, he says, “we want to help students get out of that rut.”
Khanmigo was tested in schools across the country last academic year with about 100,000 students and teachers, which helped detect bot hallucinations and provided DiCerbo and her team with a wealth of recordings of conversations between students and bots for analysis.
“We’re looking for things like summaries, guidance, and encouragement,” he explains.
It’s not yet known how much Khanmigo has closed the AI engagement gap. Khan Academy plans to release some summary data on student-bot interactions later this summer, according to DiCerbo. Plans for outside researchers to assess the tutor’s impact on learning will take longer.
AI feedback works both ways
Since 2021, nonprofit Saga Education has also been experimenting with AI feedback to lend a hand tutors better engage and motivate students. Working with researchers at the University of Memphis and the University of Colorado, Saga team pilots in 2023 fed transcripts of their math tutoring sessions into an AI model trained to recognize when a tutor prompted students to explain their reasoning, refine an answer, or initiate a deeper discussion. The AI analyzed how often each tutor took these steps.
By monitoring about 2,300 tutoring sessions over several weeks, they found that tutors whose coaches used AI feedback peppered their sessions with significantly more of these cues to encourage students to engage.
While Saga is considering the possibility of AI providing direct feedback to teachers, the company is cautious about doing so because, as Brent Milne, vice president of product research and development at Saga Education, says, “having a human coach at the center is really valuable to us.”
Experts expect AI’s role in education to grow, and its interactions to become increasingly human. Earlier this year OpenAI and startup Hume AI separately shot “Emotionally intelligent” AI that analyzes tone of voice and facial expressions to infer a user’s mood and respond with measured “empathy.” Still, even emotionally smart AI likely won’t meet the challenges of engaging college students, according to Brown University computer science professor Michael Littman, who is also director of the National Science Foundation’s Division of Information and Knowledgeable Systems.
No matter how human the conversation is, he says, students understand at a basic level that AI doesn’t really care what they have to say in their papers or whether they pass or fail their classes. Students, on the other hand, will never really care about a bot or what it thinks. June study in the journal Learning and Instruction found that AI can already provide decent feedback on student essays. It’s unclear whether student writers will put in the effort and care to do so, rather than hand the task off to a bot, if AI becomes the primary recipient of their work.
“The human connection is so important in learning,” Littman says, “and when you just take people out of the equation, something is lost.”
This story about I have teachers was produced by The Hechinger Report, an independent, nonprofit news organization focused on inequality and innovation in education. Sign up to receive Hechinger’s Bulletin.