The internet is all a-flutter with worries and speculation about what ChatGPT and AI in general might do to K-12 (and we are no exception). While it seems likely that AI will reshape the way teaching materials are written, there is even speculation that it could “substitute teach”.
But for some time, AI has had real promise in early childhood education (ECE), teaching children robotics, computer related skills, and reading. The robotics and computer skills make sense, but literacy is an interesting inclusion. The research is proving promising and investment is following.
Theoretically, there are two main benefits. First, it would be quicker to detect when a child is struggling with a particular word or phrase and be able to help with a pronunciation guide. Second, is that it can be customized to meet the needs of different learners. Since both of these things happen almost naturally when a child is being taught by an engaged teacher or parent, it seems unlikely that this is the full reason. But the thing that AI would have is unlimited time and, perhaps more importantly, unlimited patience.
Some of the early research into ECE and AI was DARPA’s Machine Common Sense (MCS) program, which “aims to create a service that learns from experience, like a child, to construct computational models that mimic the core domains of child cognition for objects (intuitive physics), agents (intentional actors), and places (spatial navigation).” DARPA used what they have learned to program robots and other technology.
This line of research has led to a lot of interesting results on AI in ECE. One of the resulting questions is why it is effective in ECE when humans doing chat based tutoring isn’t that useful in K-12? That seems to still be up in the air but it could be a result of a petri dish effect. AI might be good at teaching under lab conditions, but underused, just as Paper Tutoris is, if a platform were deployed at scale.
But that isn’t stopping anyone from trying. Amira Learning was a startup darling when it launched in 2019 and secured $5M from a few VC’s (including the now giant-er HMH) and it gained more steam (and funding) in 2021 when they made their play for post-COVID classrooms.
But Amira is not alone. Readability Tutor has a similar pitch for what it wants to offer. And, although this is far from a perfect measure of quality, markedly better app-store reviews than Amira.
But none of that provides a complete answer for why AI works better in ECE than in other places, but it could be down to money. In spite of the fact that education early in a child’s life is perhaps the most important thing that we can spend money on, ECE teachers are paid poverty wages and as a result are thin on the ground. Because of this, the teaching shortage is dire in pre-school education.
But it also could be that very young minds exposed to AI could both learn and teach. As the Hole in the Wall Project proved more than 20 years ago, the children will figure things out faster than expected and may, in some cases, teach the machine. It is even true that the concepts being taught in ECE are simpler and might be more in line with what AI can grasp.
There is perhaps no period more important for learning or more useful for breaking cycles of poverty or low educational achievement that we sometimes believe the system is doing, than early childhood. And an AI tutor is certainly a lot better than nothing. In fact it is being pitched exactly that way.