Suppose that the objective of teaching writing to elite college students is to get them to write at the 90th percentile of the population. And suppose that at the moment AI can only write at the 70th percentile. This suggests that we should continue to teach writing the way that we always have.
But suppose that in a few years AI will be writing at the 95th percentile. At that point, it is going to be really hard for humans to write superbly without the assistance of AI. The process of writing will be a lot more like the process of editing. The way that we teach it will have to change.
An Analogy: The Life and Death of the Human Othello Player
In the 1980s, I played the board game Othello (a strategy game, sort of like chess or Go) in tournaments. At my best, I was one of the top four players in the United States.
Also in the 1980s, personal computing was dominated by Intel chips, which got better just about every year. They progressed from the 286 to the 386 to the 486 to the Pentium.
A fellow named David Parsons wrote a computer program for the PC to play Othello. When I first tried it out on my PC, which probably had a 286, the computer could only play in a reasonable amount of time if I had it look ahead less than 5 moves. At that look-ahead, as long as I was concentrating I could beat it fairly easily. But with a later generation chip, using the same program, I could set the look-ahead to 7 moves, and then I could almost never win.
I soon realized that the tournament Othello would never be the same. Computers would become the best players. As humans, we would have to memorize opening sequences created by computers. We might tinker with a computer’s “opening book” by changing the probabilities with which it selects different openings. In order to actually play as humans ourselves, we would have to try to train ourselves to calculate the endgame as perfectly as a computer could calculate it. I soon gave up tournament Othello. Why bother?
The moral of the story is that when the computer’s skill gets within the range of a competent human, watch out! Another iteration of improvement and the computer is going to zoom past the human.
That is a scenario that I could see playing out in the case of writing. Maybe today AI writing is “slop.” Maybe it will even stay that way. But a more likely scenario is that AI writing will improve to the point where it changes the game. Teaching someone to write at the 90th percentile when AI is writing at the 95th will seem pointless. Why bother?
From Writing to Editing
My outlook for the future of writing is that it will look more like editing. The AI will do most of the work.
A good editor can suggest topics. He can provide general guidance. He can recommend style. When the author submits a draft, the editor can propose revisions.
My guess is that in another year or two only a very few students will be able to develop enough skills to write better than an AI. But maybe every student can learn to work with an AI as an editor.
The objective of teaching writing in many disciplines is to get the students to practice thinking through an original argument at length and with precision. I think there's a level of general smartness that can only be achieved by people who have this skill. If you are just talking through an idea, that lends itself to a shallower understanding.
It may be that people won't need this sort of smartness anymore, but I'm not sure that is plausible.
“My outlook for the future of writing is that it will look more like editing.“
I think of the ability to write well as a prerequisite to being able to edit well since editing is merely the process of revising some initial draft to improve it. In other words, if you can’t write well, then you aren’t likely to be a very good editor. And, since writing well takes a lot of practice to perfect, I would expect the art of writing to continue on indefinitely just like multiplication tables or phonics.