Adaptive Learning Posts

I've written previously about the technical definition of adaptive learning and about why most platforms don't meet it. This post applies that framework to the three biggest brands in the space — Pearson, Coursera, and Khan Academy — and looks at what each actually ships in production. The conclusion isn't subtle. None of them clears the technical bar. The reasons are different in each case, and they're worth understanding because they predict where the category goes next.

Cross-domain transfer is one of the oldest open problems in machine learning. It shows up in the literature under a dozen names — transfer learning, multi-task learning, domain adaptation, meta-learning, foundation models, representation transfer — and it's been the subject of thousands of papers since the early 1990s. Hinton wrote about it, LeCun wrote about it, Bengio wrote about it, every major ML lab has multiple research lines dedicated to it. The general problem is well-trodden academic territory.

I want to spend a few thousand words pulling the word "adaptive" apart, because it's been so thoroughly abused by edtech marketing that the category has become functionally meaningless. Every platform with a quiz engine and a question bank now claims to be adaptive. Some of them have a difficulty slider that nudges easier or harder based on recent correctness. A handful do a little Bayesian Knowledge Tracing under the hood. One or two have actually built something that meets the technical definition. The rest are autocomplete with confetti animations.

Adaptive learning is a category that gets used loosely. Every edtech vendor with a quiz engine and a difficulty slider claims to be "adaptive." Most of them are not, in the sense that matters. The question that separates real adaptive learning from marketing-deck adaptive learning is a latency question, and the threshold is roughly two milliseconds.