Teaching Taste, Testing Taste
What do you do a lot of in your job?
There are lots of facetious answers that can be given in response to that question, and that would be a fun way to spend a Friday. But jokes apart, what I mean is, what is a task that your job necessarily entails? A task that you have done hundreds, if not thousands of times? So often that the task is like breathing for you - you can literally do it without thinking about it.
So often have you done it, in fact, that you know what the output of that task should look like. You can feel the output, you can know it by just glancing at it. You know, in your bones, when it is off by even a little bit, regardless of whether you have done it or somebody else. In fact, it couldn't possibly have been you, because then it wouldn't be off by even a little bit, now would it?
If you are curious about what the answer is in my case, it is judging if a class has understood a topic that has been explained. I can pick up on cues that helps me understand when a topic is "going well", when it is flying high up over the heads of the audience, and when it is somewhere in the middle. Incidentally, this is why I hate teaching online - regardless of whether the cameras are on or off, this is all but impossible with an online audience. But anyway, my answer to my own question is my ability to "read the class".
And you'd have that ability too, if you ended up teaching as many hours as I have taught. This ability of mine isn't anything special, it has come out of hours of practice. And my first point is that you have such talents too, related to your job. You too, have performed some task day after day, month after month, year after year, to the point that you are now a master at it.
My second point is that you are now such a master of this task that judging how well this task has been done is now second nature to you. I, for example, can pick on whether a class being taught by somebody else is going well or not by simply glancing inside. You too, can similarly spot whether a task at which you are a master has been done well or not. Expertise acquired through years of practice.
Let's talk about taste today.
Not in terms of the word's meaning when it comes to the culinary world, though. Taste, instead, as the ability to judge something as being "good". How do you know that something - anything - is good?
When you hear a song for the first time, or look upon a piece of art, or watch a movie - how do you know that what you're looking at is good?
When you watch a powerpoint presentation, when you look upon someone else's spreadsheet, when you read an executive summary drafted by somebody else - how do you know that what you're looking at is good?
Well, there are two ways of getting good at developing taste. One can be by looking at hundreds, maybe thousands, of versions of the output of the task, and learning, over time, about which ones are good, and which ones are lacking along some dimensions.
You can replace the word looking with "hearing", or "tasting", or some other, more applicable verb. But whatever you are doing, do a lot of it. A helluva lot. Do it until you can begin to discern nuances that very few other people can. The more you do this, and the more rigorously you do it, the better you get at judging whether something is "good"or not.
There is one important downside to this method of gaining mastery, however. You will learn, it is true, to judge if a task has been executed well. You will know, by looking at the output, if it is "good" or not. But because you have gained mastery over only the judgment of the final output, and not the process that generates the output, you may not be able to give meaningful advice about how to improve upon the output.
Below is one of my favorite passages from one of my favorite books from last year, A Gentleman in Moscow (no spoilers):
At one in the morning, the conspirators took their seats. On the table before them were a single candle, a loaf of bread, a bottle of rosé, and three bowls of bouillabaisse.
After exchanging a glance, the three men dipped their spoons into the stew in unison, but for Emile, the gesture was a sleight of hand. For when Andrey and the Count raised their spoons to their mouths, Emile let his hover above his bowl— intent upon studying his friends’ expressions at the very first taste.
Fully aware that he was being watched, the Count closed his eyes to attend more closely to his impressions. How to describe it? One first tastes the broth—that simmered distillation of fish bones, fennel, and tomatoes, with their hearty suggestions of Provence. One then savors the tender flakes of haddock and the briny resilience of the mussels, which have been purchased on the docks from the fisherman. One marvels at the boldness of the oranges arriving from Spain and the absinthe poured in the taverns. And all of these various impressions are somehow collected, composed, and brightened by the saffron—that essence of summer sun which, having been harvested in the hills of Greece and packed by mule to Athens, has been sailed across the Mediterranean in a felucca. In other words, with the very first spoonful one finds oneself transported to the port of Marseille—where the streets teem with sailors, thieves, and madonnas, with sunlight and summer, with languages and life.
The Count opened his eyes.
“Magnifique,” he said.
Andrey, who had put down his spoon, brought his elegant hands together in a respectful show of silent applause.
Beaming, the chef bowed to his friends and then joined them in their long-awaited meal.
Towles, Amor. A Gentleman in Moscow: The worldwide bestseller, now a major TV Series starring Ewan McGregor (pp. 221-222). (Function). Kindle Edition.
Of the three people who partook of that bouillabaisse, two were experts in the tasting of it. And having tasted it, they approved of it.
Only one - Emile (the chef) - was an expert in both tasting it, and in making it.
And that's the other way to acquire mastery. Not by looking at hundreds, or thousands of versions of the output, but by making hundreds, or thousands of versions of the output.
The challenge of higher education in the age of AI is three-fold. We need to teach our students how to make the bouillabaisse. We need to teach our students how to appreciate a good bouillabaisse made by something or somebody else. And we need to teach our students how to adjust a decent bouillabaisse in order to make it *just* so. Good enough to earn a chef's kiss, if you will.
That bouillabaisse, obviously, is only a metaphor. It can be a spreadsheet, it can be a report, it can be code, or it can be anything else, and it can be related to any subject under the sun.
But to go back to the metaphor, the point is to help everybody know what a good bouillabaisse tastes like, and for some of us to be really good at making a good bouillabaisse.
Right now, students are just figuring out how to generate passable versions of the soup, and even that they are doing on their own initiative, not because it is required of them in college.
No, in college, you are required to memorize and write out the recipe for bouillabaisse. By hand. And in the exact order in which it has been written in the official How To Make Bouillabaisse textbook.
An extra pinch of saffron, however, is encouraged. Any chef will tell you that.
But metaphors and bowls of bouillabaisse can take us only so far.
Here's some plain talk: if you are a student in college today, please, I beg of you, think very hard about how to develop taste for what you hope to do in your job.
Develop taste for writing code if you hope to get into, say, data analytics. Know what good code looks like by writing a lot of it. Go through the pain of having to correct badly written code. Go through the pain of having to rewrite entire stretches of it because you got some part of the plan fundamentally wrong. Go through the pain of having to start all over again, and do all of this many, many times.
Do it often enough so that judging code that you're looking at becomes second nature. Second nature to the extent that you can judge code by looking at it, by developing a feel for it. And expert enough so that you know how to correct it, or you know how to talk to an LLM so that it can correct it.
It need not be coding, it can be report writing. It can be financial analysis. It can be project management. It can be anything in any field.
The least important part of your CV from here on in is your degree. The least important part of your job application is your CV. The least important way to get a job is by submitting a job application. The amount and the quality of your work is what matters, and the latter is a direct outcome of the former.
Quality is nothing but taste, and taste is what is left when you improve upon avoiding mistakes.
So if you are wondering how to think about studying in the age of AI, go ahead and do a lot of whatever it is that you hope to do in your career. Don't worry about how bad your first output is. Worry about making your second one a little bit better, and repeat. Publicly.
Until somebody, somewhere goes Magnifique.
But if you are a student, that is my sincere advice to you: make a helluva lot of batches of your bouillabaisse while you are in college.