What chess can teach us about education and AI

The computer program Deep Blue beat chess world champion Gary Kasparov in 1997. For 25+ years computers have been better than the best humans at chess. Despite this there are some fascinating trends with humans and the game. The first, is that the top human players are in aggregate better than they have ever been before. The second is that the game is more popular than it has ever been despite the dominance of machine learning. I think there is something to be learned from these trends in chess in relation to our collective interest in teaching and learning. 

First, just like in chess, AI can make us better at what we do. Any chess game can now be analyzed by any number of programs like Stockfish to help players understand where we made a blunder or an alternative line of attack. In terms of AI and academics, we have already started to see this play out in key scientific fields like protein folding where human expertise and judgment combine with powerful machine learning to solve problems on a scale we could not have imagined 15 years ago. AI is good at performing tasks, it is not (currently) good at doing jobs. Tasks are a big part of how we promote and evaluate learning. The solutions vary by discipline, but there is an opportunity to build on the inspirational model of chess and think about how AI can make us and our students better at the exploration and discovery most of us fell in love with as teachers and researchers. 

Second, just like in chess, the existence of a computer that can complete a task does not mean humans are no longer interested in the task. For most people who play chess the goal is not to be the world’s most effective player (machine or human), but to engage in the process, get a little better, and lose yourself in a competitive puzzle of sorts. We incentivize the completion of tasks with grades, but it is also possible in some circumstances students find value and meaning in the process–it is also possible we need to reframe how we explain why we are doing tasks to students. There already exists an edge case: a student who truly loves the material so they will already do the work themselves. How do we inspire a larger fraction of a class to engage in the process because the process is valuable to them? This is difficult to imagine, but the example of chess shows us it is possible. 

The next phase of teaching and learning is opaque. I think the example of chess is an interesting one because it points towards a historic instance where machines have long surpassed humans, and yet our human interest in the activity has expanded. There are of course differences. Most people play games for fun and have a different disposition towards a 5-paragraph essay or a math proof. The point is not that these situations are completely analogous, but that we know humans remain in activities machines can do better. The challenge would seem to be how to make those things compelling enough to keep people interested.