The idea that using AI for educational tasks is like taking a forklift to the gym has been around for at least a couple years. AI art has been compared to wanting congratulations for driving a marathon. I think especially in an educational setting it's important for students and faculty to think about what they're getting out of it, what's the purpose, and how to align AI use (or non-use) with those goals. It's also valuable to remember, though, that you're not always at the gym, and sometimes a forklift is the right tool for the job.
The analogies suggest that there are different types of tasks with different expectations for their accomplishment.
- Effort is the whole point tasks
- Outcome is the evidence tasks
- Outcome is the point tasks
Effort is the whole point #
This is the forklift in the gym tasks. No outcome here matters to anybody. What does matter is that the effort was made, and through that effort, the effort giver is transformed in some way.
Education largely falls under this category. The point of practicing multiplication, writing a paper, or balancing chemical equations is to put in repeated effort to ingrain concepts into the way you think about some part of the world. You are solving solved problems, but in doing so you are exercising your abilities and improving your understanding of how something works.
In these tasks, the "gainz" only come from your effort, and can't really be gotten any other way. Skipping the effort with AI means you have learned nothing and contributed nothing.
The outcome is the point #
I'm currently building a website to track my kids money and to encourage them to save money for things they actually want rather than buying whatever happens to catch their attention when they have $5 to spend.
For a website like this, nobody cares how it got made. In the same way I don't care if you used Rust or Elixir or NodeJS to build your website, I don't care if you built it solo by hand or with a team of 20 people or with an AI agent. I care what the website can do, if it meets my needs, and if it does so reliably. Now, I say nobody cares how it's made, but that's not quite true, either. I wouldn't call myself an AI hater, but I would say I'm AI-skeptical. I'm also skeptical of people who don't know what they're doing who use AI to build things far beyond their understanding. The risk of data loss and security vulnerabilities is concerning, even if you ask Claude nicely to write secure code.
Sometimes a forklift is the right tool, and sometimes full-blown AI is fine. Outside the classroom and working on a side project? Go nuts.
Effort imbues meaning #
Then there's the third category - where the output matters, but it was the effort used to generate the output that makes it matter.
Art falls into this category. Unlike educational tasks where the effort is the point, in this case, the outcome is evidence of human effort, and it is the effort that gives the output meaning. This is why people sneer at AI art; our conception of art is that it is the embodiment of some part of the soul of the artist. It was an original vision of the artist come to existence through human effort. Without that effort, the value is lost even if fundamentally the output is the same.
We found a similar effect in a recently-submitted paper. For certain tasks like writing a love poem to your spouse, we care about the output, but even more we like knowing that a person sat down and put thought into making it. Learning that the output came from a thoughtless AI prompt instantly devalues it.
Go to the gym! #
In the same way that a sedentary lifestyle threatens our physical health, AI poses an existential threat to our ability to think critically, reason carefully, brainstorm creatively, and communicate effectively. I don't think those are uniquely human abilities, but they are necessary human abilities that, if not exercised, will atrophy in the same way muscle and lung capacity fade with disuse.
It's not hopeless, though. We have gyms and personal trainers and exercise regimens to exercise our muscles. We can do the same for our critical mental capabilities.
- Practice writing without AI. For me, that's this blog. For you maybe it's one part of your job that you leave AI out of. Maybe it's a journal/diary. My personal rule is if the writing is primarily meant to be actually read by people, it should be written by me. When I write here, it's me. When I review a paper, that's me, too (though I might use AI evaluate my review for correctness, I write every word).
- Use the time you save with AI to stretch yourself. While your coding agents are cranking away at a problem, read something relevant to your field. While Cowork polishes your presentation slides, close your eyes and think, plan, or brainstorm ideas. For something, anything.
- Just think about it. If you're using AI for more things throughout the day, think about what parts of your mind you're no longer exercising, and decide if that's something you want to give up, or if it's something you can practice a different way.
- I'd love to hear ideas for how to stretch your brain in new ways.
Humanity is the point #
I think what it really comes down to is that humanity is the point. Of the whole thing. Effort matters because of how it teaches and changes a person. Art matters because it was created by people. The websites I use matter because of how they help me do what I want to do.
Build things for people, write things for people, read things that people wrote. And for heaven's sake go to the gym.