Why study? ChatGPT can do this for me
human-writtenⓘ
A few days ago, one of my kids got a bad grade on a French test, and he went something like “why do I have to study anyway? ChatGPT is able to do this for me anyways”.
I could have responded on the spot with platitudes like:
- “Because you need a diploma to get a good job.” A: you don’t. And B: a good job is a very bad reason to go to school.
- “Because you won’t always have ChatGPT in your pocket.” We said the same about calculators, and boy, did that turn out to be wrong.
But instead, I froze.
It was a very good question, and very good questions require a thoughtful response.
So here is my thoughtful response.
Toil vs Struggle
Unrelatedly, a few days later I stumbled on this post by Kent Beck, in which he makes a distinction between toil and struggle.
Toil is hard work that only results in, well, the result. His example is reformatting a 2013 essay: stuff you need to do, but won’t get any satisfaction out of.
Struggle, on the other hand, is hard work that teaches you something new. His example: a math proof he worked on for 3 hours.
He finishes his post with “I’m delighted to have the genie bear the toil. Leave me my struggle. I need it.”
Training vs Racing
I love how Kent Beck sorts work into two piles: work you can learn from, and work you can’t.
However, I do believe the difference between the two piles is a choice.
You can always choose to learn at least something, even with the most mundane work. When I’m trimming my hedge - something I do twice a year, the same hedge - I usually don’t make that choice, but I could choose to learn how to do a better/faster/cleaner trimming job.
So I’d put on my Stoic hat and say that it’s your attitude towards the work that matters. Essentially: do you choose to race today, and just get things done? Or do you choose to train, take a detour and struggle?
When you race, e.g. in an actual marathon, the only thing that matters is getting to the finish line as fast and as efficiently as possible. Training for one, however, is in many ways the opposite (e.g. long runs are usually as slow as possible, and never the full distance).
Work is a race
Now, as you go from elementary school to middle school, university and the workplace, you slide more and more along the scale from training to racing.
But paradoxically, the older you get, the more you start to value training over racing. (I am also fully aware that I sound like an old fart, typing this.)
The definition of “work” is - in my book - to get shit done: set a goal, draw a line as straight as possible, get there as efficiently as possible. Anything that takes your eyes off the prize (ugh, learning) costs time.
It’s a cost, quite literally. But you could look at it as an investment too. Because the issue with just getting shit done is that next time you encounter a similar situation, you do the same things all over again. Which is not very, dare I say it, efficient.
Unfortunately, you can only tell after the fact whether it was a cost or an investment. That’s the nature of learning stuff: there is no straight path.
At work you race, so you let the agent do what gets the job done. As quickly as possible. However, a racer who never trains gets weaker, so it’s important to strike a balance: make time to train.
An example is this blog. In Writing with AI I said I’d write posts by hand “as much as possible”. Which is what I’m doing right now: a ✍️ post is training, an 🤖 post is racing. The labels on each post make that visible, and by making it visible, I force myself to train :)
School should be training
Learning, progress: that’s ultimately the goal of education.
But school seems to be more about racing than training: most (not all!) teachers seem preoccupied with stuffing knowledge into our kids’ heads, and with teaching them how to solve specific types of problems. Kids are taught how to use Excel and PowerPoint, not how a network or an LLM actually works.
Creating a PowerPoint is clearly on the getting-shit-done pile. Exactly the kind of thing ChatGPT is great at! At school, it’s called cheating. At work, it’s called efficiency.
In my humble opinion, an education system should focus on teaching problem solving. The subjects (French, math, computer science, …) are nothing but a means to an end. Because we want kids to not just solve today’s problems, but also tomorrow’s, which nobody knows yet.
My answer
My answer to the question “why do I have to study anyway? ChatGPT is able to do this for me anyways” is very simple: ChatGPT can run the race for you, but it can’t train for you. And there’s a lot of value, and fun!, in training.