Learning takes time

I recently found a sticky note from around 2004 on which I had written: “In the future, artificial intelligence may obviate the need for certain skills, but the act of learning is a very human process that proceeds at human pace.”

Since then I’ve been puzzling over what exactly I meant by that. (It’s a classic example of obvious or profound?) Here’s what I think I meant: The rate at which humans learn is essentially limited. Of course, the pace of learning does vary somewhat in different situations, but this variation tends to be within, say, an order of magnitude and there is no known way to get around these limits.

Whether by impatience or optimism, I frequently forget this. As soon as I learn something useful or exciting, I start trying to rapidly explain it to others, expecting them to learn it using far less time and effort than it took me. While in some sense it’s generous to assume that others are such fast learners, it’s also a form of hubris for me to believe that my explanation is somehow orders of magnitude better than the explanation given to me. It would be as if I somehow found a magical shortcut that makes teaching and learning easy.

None of this is to say that great teaching doesn’t exist — it does, and in every domain teachers have worked tirelessly towards making learning more efficient and more fun. Motivation, support, curriculum, collaboration, and many other factors help optimize human learning. The point is simply that these optimizations have diminishing returns. That is, there’s a theoretical maximum implying that it’s impossible to learn all of calculus in one week. (On the other hand, it is possible to teach in a way that is arbitrarily ineffective — say, by putting a student in a room with no supplies and yelling at them all day.)

If all of this is true, then we would not expect any educational technology to dramatically improve student outcomes, unless we were replacing dramatically ineffective prior teaching methods. In other words, to claim that a dramatic improvement is possible is to state that the current methods are dramatically poor — in which case, there are probably plenty of alternative improvements which do not require advanced technology.

The only way technology can dramatically “speed up learning” is by making certain skills obsolete, so that it is no longer necessary to learn them at all. For example, humans in the developed world can get along fine without knowing how to grow food or survive in the wilderness (or do long division). Modern technologies have obviated the need for those skills. Some people might still want to learn them, but for those who don’t, the learning time has been effectively reduced to zero — a dramatic advance! That in turn leaves time and energy for learning other things that have become more important (say, computer skills).

This is why my focus within educational technology has always been toward obviating the need for a skill rather than tweaking the learning process or transferring existing lessons to tablets and smart boards. This is why I’m so interested in Bret Victor’s work on graphical tools for math and programming that could obviate the need to learn traditional, difficult methods for algebra, calculus, and debugging. It’s why I’ve spent so much time working on tools that make it possible to do data analysis safely without learning all of the details of statistics and visualization techniques.

But perhaps my sticky note was simply a reminder to have patience for learning. Wisdom is prized precisely because it cannot be short-circuited.

Leave a Comment