Five Interesting Moments From the Khanmigo Segment on 60 Minutes
Also: can someone help me understand what is going on with MagicSchool right now?
60 Minutes did a segment on Khan Academy and Khanmigo this Sunday. Here are five points of interest from the segment.
1. The computer vision demo.
I have little doubt that one day AI will correctly distinguish the hypotenuse of a right triangle from its height, but today was not that day. (OpenAI CEO Greg Brockman pivots quickly, asking the AI to sing a song about triangles in a British accent.)
Look, I spent my morning learning about learning in a Grade 8 math classroom and I can promise you that the students were creating diagrams that were much less structured and much messier than Khan’s precise, high-contrast drawing of a triangle. Whether computer vision can support the math I observed today, and in a form students can easily access (i.e. not their mobile phones), is a very interesting open question.
2. Asking teachers to do the impossible.
A science teacher named Melissa Higgason is profiled in the segment. At one point she pulls up a student’s list of Khanmigo chats and tells Anderson Cooper:
If I wanted to pick a specific student, I could come down here and really dive into what that student’s been looking at in Khanmigo.
“Could” is really rampant in edtech right now. People are busy dreaming possible futures for a technology that is two years old. I am interested, meanwhile, in what teachers are doing with it now, especially within the tight constraints of the job we have defined for them.
Outside of a few outliers (a couple of AI guys are making “grade the chats” their whole thing) this use case strikes me as pure science fiction. Yes, teachers could click into several chat transcripts x 35 students x 6 classes every day and do something meaningful with them. But what? And how? We should not expect them to do that given the meager preparation time we give them daily. We owe it to teachers and students to be more serious here.
3. Looking forward, not backward.
This is Sal Khan’s second feature on 60 Minutes. His first was not mentioned, nor did Cooper ask him about his predictions in that feature. In that first interview, Khan promoted a model of individualized education with video explanations and auto-graded exercises and “a room of 20 or 30 kids all working on different things.” Schools have not taken that model up in any meaningful numbers in the subsequent twelve years. Anderson Cooper could have asked Khan what he learned from his first attempted revolution in education, but he didn’t, which was certainly a kindness.
4. A couple of millionaires talking about “leveling the playing field.”
Anderson Cooper (a Vanderbilt scion) says, “If every kid could have a private tutor, that would level the playing field,” and Sal Khan (making >$1MM / year, per Khan Academy disclosures) responds, “Yeah, that’s the dream.” It should not surprise us to see economic elites dream of leveling the playing field through technology contracts that would benefit other economic elites.
Other countries level the playing field for their kids through redistributive policies like child welfare, public housing, nationalized health care, etc, all of which would require increasing the tax burden on economic elites. It should not surprise us to see which of those dreams receive fawning coverage during primetime corporate media.
5. Content and connection.
After a segment full of AI feedback—voice feedback from OpenAI, text feedback from Khanmigo—it was almost overwhelming to watch a teacher tell a kid named Ella, “Good work.” It is an entirely routine, even mundane, piece of feedback in the scope of a teacher’s day. But after all the artificiality of the rest of the segment, it felt like one person giving another person the entire world.
Feel that moment and wonder two questions with me:
What kind of emotional connection do students need to learn from a tutor?
Will AI tutors ever be able to establish that connection?
I can appreciate a range of responses here, but my skepticism about the transformative potential of AI basically boils down to “a pretty strong one for 95% of kids” and “I don’t think so” as my answers to those two questions.
At one point, Anderson Cooper draws a sketch of the human body. The AI voice says cheerfully, “It’s a good start.” Cooper says, “Don’t patronize me,” in response. The emotional connection feels, in a word, artificial.
Odds & Ends
¶ Someone please help me understand MagicSchool’s deal right now. Their founder Adeel Khan comes from schools and seems genuinely interested in what students need and teachers offer. I don’t want another edtech company to have to scrutinize right now. But they keep exaggerating themselves in very strange ways. Like popularizing a statistic about AI-generated materials that lacks any empirical basis. Like attributing results from a single teacher’s classroom to a 38,800-student district. Like claiming “love” from four million teachers on a “wall of love” that is 76% not teachers. And then earlier this week saying teachers are so using AI next to a … Google Trends graph? Like Khan knows that isn’t a graph of usage, right? But—what?—he’s counting on us not knowing?
Look folks, I want to believe as much as anybody that we’ve really truly found the One Weird Trick for human cognition, that we’re all on a rocketship ride to easier teaching and better learning. But when the leading teacher copilot tool cuts so many corners in its self-presentation, I start to think, you know, maybe I’ll just take the next rocket, thanks.
Adeel: how many of those four million teachers have logged into MagicSchool twice? Literally just twice.
¶ ElevenLabs is a voice tech company, not an edtech company, so I am giving them only half a demerit for this bummer demo of an AI tutor. Full demerits, however, to all of the people I have seen across edtech social media absolutely thrilling to what is—we need to agree on this somehow—really bad math and really bad teaching. If you’d like to read some expert-level commentary on the math and teaching of that video, check out this Bluesky thread with me and some of my pals.
AI offers all of us one great gift, now, in the present: the opportunity to reorient ourselves to principles of good teaching, good learning, and good math. Let’s not waste it.
My district is buying new AI tools on what seems like a weekly basis. I am getting surveys on what training I want on each tool. What I want is training on autism, on ADHD, on homelessness in teens, on supporting teen parents, and on how to build and maintain trust quickly with people you have just met. And, no, I don’t want to ask an AI tutor about any of it.
Humans have been doing math forever (the Babylonians wrote about the relationships between the sides of a right triangle and its hypotenuse ~4,000 years ago!). If teaching math was simply about finding the "best explanation", we'd have found most of them by now, and no kid would fail geometry in 2024.
Humans have been *learning* forever (ancient Greece had students studying reading, writing, and math ~2,500 years ago!). If education was simply about "giving teachers the right tools," we’d have discovered them by now, and no teacher would be burnt out in 2024.
But that's (obviously) not the case.
So when the loudest, most lauded voices are claiming that they've transformed education because they've discovered the best explanations and the right tools, it's hard to take those claims seriously.
The problem is that the issues of school are more fine-grained than explanations and tools, which makes them hard (but not impossible!) to address. Hard (but not impossible!) problems take time to solve - they require a degree of patience **for developers**. Instead, the burden of patience is often placed on teachers and students: "just wait for the next model to come out! But for now, here's something lesson plan-shaped."
I can't pretend that gen AI is all hooey–my gen AI use has significantly changed my professional life for the better over the past 12 months–but we also can't pretend that tools made by and for R&D folks like me can be simply plopped into a classroom setting.