The story conveyed in this article is something that I’m deeply familiar with and troubled by, but the quote near the end of the article had me wondering:
“AI might be better — it is better — at lots of things, but I have a comparative advantage when it comes to teaching because I am good at the mushy human stuff.”
What exactly is this “mushy” human stuff? The answer that I always thought I had found is that it varies tremendously, from case to case, from time to time. And more important, it is not necessarily something that people value much — even on the receiving end! There is a reason why the market for formulas and textbooks is huge: people are busy; they don’t want to take time out to think; and this is not a complaint about people’s attitudes. One critical lesson that I learned from talking to people integrating technology into military applications is that you need to make things simple enough that they can be operated by 19 year old high school grads working under fire, not trained engineers, and that means everything has to be reduced to something simple, clear, and formulaic. A classic engineering problem, really, but with an emphasis beyond just the “engineering” part — you want to not only develop an application that is easy to use, but also design a procedure for using it that does not involve generating complex questions and “funnies.”
This strikes at the point of departure that I had always seen to be separating “science” and “engineering.” Science thrives on the “funnies,” where things don’t work as expected, where the deviations from the norm are found, where the errors show up, and something interesting might be learned that we hadn’t expected. To an engineer, a “funny,” something unexpected, an “error” is not only a waste (the attitude that I found troubling among many “data science” types, but, ultimately, understood), but a dangerous threat, literally. That’s not something that you could give a 19-year old private a basic set of instructions on how to handle. You need to bringing in someone who can think things through to deal with situations like that, and not under fire either.
Of course, the difference between the “science” and “engineering” mentalities go far deeper, to the medieval distinction between liberal and mechanical arts. “Liberal” in the former description referred to the fact that its practitioners were freed from having to work with their hands for living, that they had support from sources other than their labor so that they could afford to think about things beyond their immediate and practical applicability. Mechanical arts were more like the description of adapting technology for use by 19-year old soldiers. Whatever they were, they had to work simply and reliably, as long as the instructions were followed, without having to resort to thinking beyond the manuals and formulas. (I am being a bit unkind to “engineers,” who will no doubt react by noting that this is not how they work. BUT it is the characteristics of what they produce, even if not how they work. The 19 year old operators of military gadgets are not engineers, but engineers make sure that they don’t need anything more than manuals — and the rest of how they “really” work is, well, just magic.)
There is a reason why “universities” became the home of “liberal arts,” and how, for centuries, they were reserve of the people who had other sources of support — the Catholic Church, for most of its early history, and later, the wealthy, and the nobility, as well as those who were, for whatever reason, favored by the powerful and wealthy. Their products, for the most part, were not “practical.” It was only in the 19th century that applications of advanced sciences found widespread engineering use, and that was because science had been sufficiently refined to the point that they could be reduced to a set of “formulas” that reliably captured how the universe operated — reliable enough to be taught from practical “textbooks.”
In a sense, we want to repeat the transformation of some aspects of higher education in the 19th century, of reducing the universe to reliable enough formulas that, if learned properly, could be put to practical use. Does this make sense, though, given the world today? I’ve spent enough time in social science academia to see major discipline wide epiphanies strike several fields — economics, most notably, in recent decades (and was impressed enough that economics, as a field, has matured enough that it could absorb and learn from the crises productively), but also poli sci, sociology, and psychology (I don’t know as much about the latter two, though).
The problems that these fields face is that, in terms of how much we understand the topics of our endeavor, we are still in 17th or 18th century, rather than the 19th. We know a lot of stuff, but we don’t know a lot also, and we are not yet ready to package them into reliable enough formulas suitable for practical use. I know for fact, given I was involved in that very line of research, that while the models of electoral politics and such things that we have in political science are quite good at capturing what happens under “average” circumstances, there are enough atypical events going on in some corners of the political universe all the time that it is very dangerous to rely on the standard models and formulas, so to speak, as guidebooks for the proverbial 19-year old privates. N. N. Taleb’s caution holds true in these situations: “never try to wade across a river that is, on average, 4 feet deep.” Heck, if I don’t know the distribution of the depth, I wouldn’t try wading across a river that is a foot deep on average — “average” is the number one way people lie with statistics, even to themselves. Even in high stakes businesses like high finance, lack of sufficient appreciation for the variances lead to colossal mistakes. Other fields of endeavor are even noisier and we have much less “skin in the game” for understanding these. I would imagine that this is still true for most of academic disciplines at universities, even in natural sciences or even engineering, given how things are changing rapidly in certain aspects of how the world works. We need to be able to step back and think, and, as such, more of the medieval university. Unfortunately, this is exactly the opposite of the direction that the higher education is actually taking for the most part.
PS. One might point out that the higher education is most decidedly not designed to address the needs of 19-year old privates who are operating machinery under fire but to help train engineers who are needed for dealing with more complex problems. This is indeed true and is at the heart of the problem: for various social and political reasons, higher education is being sold as a remedy for the problems facing 19-year old privates and is subject to much pressure to adjust itself to meet their needs, even if that means making the contents less appropriate for the engineers. It seems more like we need better general education system, where the needs of the 19-year old privates are better served elsewhere so that the higher education can focus on addressing the needs for the fewer engineers who need greater in-depth training. It does remain, however, whether the 19 year old privates are best served by formulaic training without being taught to “think.”
PPS. The problem, in a sense, is that the “artillery school” (or wherever else that the 19 year old private “should be” sent to learn his formulas) does not carry the same prestige as a “university,” and this lack of prestige leads directly to disparity in the social standing. This is, in a sense, a real problem, but not necessarily a natural one. One interesting (albeit a bit unusual) counterexample is the social standing of the staff and combat officers in 19th century Prussia: the aristocrats were often among the fighting officers while the middle classes were overrepresented among the general staff, reversing the usual hierarchy where the highly educated lay at the top. One might also take as example 17th century France, where the New Nobility, drawn from the technocrats who came from the humbler social strata and entered the service, in particular, of Louis XIV, increasingly entered the ranks of the aristocracy alongside the old blue bloods. But it is true that, in both cases, the new aristocrats, whether German general staff or the French technocratic nobility, swiftly retrenched themselves, perhaps even more arbitrarily than their more “genuinely” aristocratic predecessors. Is this something else that’s taking place again today?