So far, modern educational practices haven’t been able to solve Bloom’s 2 Sigma Problem, which suggests that personalized one-on-one tutoring could shift the bell curve of student achievement two whole standard deviations to the right. Could AI tutors be the missing link that could transform Bloom’s 2 Sigma Problem into a 2 Sigma Solution? Sal Khan of Khan Academy thinks so. In this engaging TED talk, Khan makes an argument that the thoughtful implementation of AI tutors could revolutionize education.
- One-on-one tutoring can have a substantial positive impact on student achievement, but practical and economic constraints limit access. AI could change that.
- Students can use AI-enhanced learning to “converse” with historical figures or fictional characters, engage in Socratic debates, and workshop their writing.
- AI is a powerful tool that requires intentional development to mitigate possible negative effects so it can help teachers as well as students.
One-on-one tutoring can have a substantial positive impact on student achievement, but practical and economic constraints limit access. AI could change that.
Back in the 1980s, educational psychologist Benjamin Bloom found that students exposed to one-on-one tutoring tended to perform much better than students in the typical classroom setup with one teacher and 30 students. In other words, one-on-one tutoring “could take your average student and turn them into an exceptional student,” and “can take your below-average student and turn them into an above-average student.” But how do you provide a knowledgeable tutor for every single student?
“We’re going to do that by giving every student on the planet an artificially intelligent but amazing personal tutor. And we’re going to give every teacher on the planet an amazing, artificially intelligent teaching assistant.”
Using the power of generative AI, Khan Academy has created Khanmigo, an AI tutor that doesn’t just spout information, but engages with students’ learning processes by asking questions, theorizing how students think and where they may have gone wrong, and guiding them to correct their mistakes.
Students can use AI-enhanced learning to “converse” with historical figures or fictional characters, engage in Socratic debates, and workshop their writing.
Since generative AI has become available to the public, teachers and journalists have raised concerns that students will use it to cheat. If they can just ask a computer for the answer, won’t they fail to learn academic skills for themselves? With six months of fine-tuning, the Khan Academy team was able to modify ChatGPT to make it a more suitable tutor. Rather than simply giving students the answers, they trained Khanmigo to “think before it speaks,” scanning the student’s work to find the error in his or her thinking. Khanmigo then engages students in Socratic debates and directs them to find answers themselves.
“Not only does it notice the mistake, it asks the student to explain their reasoning. It’s actually doing what I would say, not just even an average tutor would do, but what an excellent tutor would do.”
Many people are concerned with the effect generative AI will have when it comes to writing, worrying that students won’t learn to write on their own. But with the right guidance, “AI doesn’t write for you, it writes with you.” Khanmigo can help create an outline, and once students have finished their essays, Khanmigo can give them personalized feedback to improve their writing. Smaller kids can also play writing games in which they construct a story alongside Khanmigo.
There are other features that make using Khanmigo a “magical” experience. While studying The Great Gatsby, one student was puzzled by Jay Gatsby’s obsession with the green light shining from Daisy’s pier across the water. Dissatisfied with the explanations she found online, the student asked Khanmigo to explain it to her. The AI tutor responded in the voice of Jay Gatsby, and the effect was so convincing that the student apologized for taking up Gatsby’s time at the end of the conversation.
AI is a powerful tool that requires intentional development to mitigate possible negative effects so it can help teachers as well as students.
As with many new technologies, AI is controversial. Fearing its use by bad actors to create a dystopian future, some people suggest that experimentation with AI should slow down or stop entirely. Unfortunately, the best way for bad actors to seize control is for good people to renounce AI.
“What’s really going to happen is the rule followers might pause, might slow down, but the rule breakers, as Alexandr [Wang] mentioned, the totalitarian governments, the criminal organizations, they’re only going to accelerate.”
AI provides learners with educational tools that might otherwise be inaccessible, and has potential to help teachers with lesson plans and grades, freeing them to spend more time and energy on their actual students, but these are only a few of the potential positive use cases. As long as ethical people fight for positive use cases, implementing thoughtful guidelines and sensible regulation, AI “can be used to enhance HI, human intelligence, human potential and human purpose.”
About the Speaker
Sal Khan founded Khan Academy, a nonprofit educational organization with a mission to provide free education to students around the world. Khan holds degrees from MIT and Harvard Business School, and worked as a hedge fund analyst before founding Khan Academy.