Sal Khan Still Clueless

Matt Barnum's piece for Chalkbeat in which Sal Khan unsuccessfully reflects on his latest education misfire, has opened the door for plenty of folks to take a swipe at Khan and his consistently bad ideas, but lordy, the man deserves all of it. Probably only Bill Gates has managed to be so consistently wrong about education and yet so often hailed as a visionary thought-leading wizard of golly-whiz-bangery. 

Khan deserves to be examined and dissected every time he pokes his head out, because he so consistently embodies what the Ed Tech Overlords get wrong, over and over and over and over and over again.

Sometimes he gets so close. Early on, Barnum offers this quote about the flop of Khanmigo, the Khan Academy chatbot that was supposed to be a tutor:

“For a lot of students, it was a non-event,” Khan told me recently about his eponymous chatbot, Khanmigo. “They just didn’t use it much.”

Good first step. Lord knows that lots of ed tech salesmen didn't even get this far in looking at the fate of their shiny product. But Khan can't get to the crucial next step, which is to make a serious inquiry into why his targeted customers didn't want to use his product.

Barnum, God bless him, knows the issues here:

Khan’s comments are an acknowledgement that AI has not quickly allowed for the creation of an effective super-tutor, as some initially hoped. It’s an early indication of the limits of AI to drive massive learning gains, long an unrealized goal of various technologies. While Khan remains optimistic about various uses of AI in education, he’s also come to see its limits.

But rather than examine the question, or the assumptions behind his use case, Khan indulges in another classic ed tech behavior-- moving the goalposts.

“I just view it as part of the solution; I don’t view it as the end-all and be-all,” says Khan, the guy who literally wrote a book about how AI "will revolutionize education" thereby being a be-all and end-all. 

It is, as Audrey Watters has laid out a zillion times, fundamental to ed tech wizardry to be fully ignorant of the history of failed ed tech. Khan launched Khanmigo citing Benjamin Bloom's 1984 essay about "Two Sigma" tutoring, a magical kind of tutoring that would leapfrog student learning, except that folks like Khan who like to trot this out don't seem to have looked at it very carefully. Education Next ran a new piece by Paul T. von Hippel breaks it down effectively. I recommend it as an antidote to folks trying to sell you 2 Sigma tutoring, but here are just a few facts about the article that Khan skimmed over. The chart that Khan and others like to use to illustrate Bloom's "findings" is not an illustration of actual data, but Bloom's hard-drawn illustration of "this is what it would look like." The actual data comes from work of two grad students who worked with a small number of subjects, tutoring them to prepare for a very narrow, specific test. Oh-- and the tutors were humans. 

But ed tech whiz guys don't just ignore the past. They also are committed to ignoring evidence from the present. It was Barnum who went back to one of the early adopters of Khanmigo to see how that was going. It's not. She found "that students didn't really care for the bot" and that "If students don’t engage with the material enough to know what they’re looking for, then an AI like Khanmigo doesn’t necessarily help." In a really key observation, she notes that "there's more enthusiasm for the product among administrators than teachers."

When ed tech overlords do encounter evidence that their shiny thing is not working out well, they inevitably fall back on one of two explanations. Kristin DiCerbo is the Chief Learning Officer for Khan with zero hours in a classroom-- she worked two years as an elementary school psychologist, then worked for Cisco, Pearson, and now Khan Academy. She explains to Barnum that Khanmigo can only respond to what students ask, and students are bad at asking questions. This time, nobody tried to blame teachers as well. This is the ed tech pitch since the dawn of time-- "This tool is great, but you need to change yourself to fit how it works rather than expecting it to fit the way you work." We have redesigned hammers, and these will be great hammers if carpenters will just replace their hands with spiral bungee cords.

Khan is disappointed that students don't seek out more help from Khanmigo, but doesn't seem to reach the obvious conclusion that students do not find it useful, and so cannot move on to the question, "How could we make this more useful?" Instead, they are just going to try force feeding Khanmigo to students by incorporating it in Khan Academy. It's the same AI business model that keeps shoving AI into every app and software that we use, like someone who thinks that if they keep setting pieces of cold liver all around the house in hopes that the kids will finally pick it up and eat it and love it. 

Ignorant of the past, ignoring the present, and dedicated to making the products they want to make first and trying to figure out what use the product could be to actual humans second-- it's the same old ed tech hustle, even for the rich and famous.