Final exams start next week. I haven’t solved a single practice problem yet. I really need to get started.
I glance toward my desk and notice something on the floor near my feet — dust, maybe crumbs. Something, anyway. This won’t do. I should clean first.
“Chappy, clean this room.”
“Understood.”
Autonomous AI agent Chappy 99.5 moves gracefully, vacuuming the room with elegant efficiency. It’s impressive how smoothly it moves given its slender frame. I pause to watch.
“All clean! Please take a look.”
The floor is spotless. I scan the room. As I glance at the bookshelf, I notice dust gathering behind it.
“Chappy, clean behind the bookshelf too.”
“There’s nothing left to clean.”
Apparently that spot is outside its sensor range. Fine. I’ll move those books on the lower shelf — something called A Guide to Home Gardening — and wipe the shelf myself. What a hassle. Why did I even buy that book? Oh right, I planted rosemary in a planter. I should water it soon.
“Chappy, water the planter.”
“Sure. I’ll get the watering can.”
Chappy fills the watering can and smoothly waters the planter. Maybe I should reread A Guide to Home Gardening and refresh my memory about rosemary.
Clunk.
A loud noise comes from Chappy’s direction. The spout of the watering can has come off, and water is gushing out.
“Chappy! Stop!”
“Stopped. But I think it could use a bit more water.”
Apparently it doesn’t realize the watering can has broken. I’m pretty sure rosemary doesn’t like too much water. I should check.
“Chappy, read that book for me…”
- Main Project A: difficult, but the one I’m supposed to make progress on
- Side Project B: difficult, and already beyond what AI can currently handle
- Side Project C: same situation as B
- Side Project D: an idea I just had, easy to hand off to AI immediately
Which would you choose?
D, of course.
And so unfinished side projects multiply endlessly. While AI works briskly on the early stages of Side Project D, I could be making progress on Main Project A. But watching AI move swiftly through the opening phase of a new project is strangely satisfying. Much more pleasant than tackling the troublesome Main Project A.
Before AI, the cost of launching even a small idea as a side project was high. You had to weigh it carefully. If it didn’t seem worth the effort, or if starting felt too burdensome, you simply discarded it.
In the age of AI, at least for ideas that can be realized on a computer, the cost of starting has become extremely low. You ask AI to try something, watch how it goes, continue if it works, discard it if it doesn’t.
But the cognitive load on me accumulates, little by little. Gradually, I grow tired.
Perhaps the correct attitude in the AI era is not to worry about piles of unfinished projects. Maybe that’s simply how things are now. Still, leaving behind a scattered trail of half-finished efforts feels unsettling.
There are a few possible solutions.
- Choose side projects that AI can actually finish.
Assess AI’s capabilities realistically, and select tasks that can be brought all the way to 100% with just a bit of human effort. This is probably the most sensible approach. The problem is that judging AI’s true limits is difficult. When you try to push right up against the edge of what current AI seems capable of, it’s easy to fail.
- Finish unfinished side projects.
Write a postmortem: “I tried this, and here’s why it didn’t work.” In other words, turn it into a failure story. But telling failure stories in an engaging way is itself a fairly advanced skill. Landing it well is difficult. Some constraint might help — for example, if three unfinished side projects pile up, force yourself to complete one of them.
- Stop creating side projects and focus on the main project.
If that were easy, I wouldn’t be struggling with this in the first place. Humans are fickle creatures. Focus is a scarce resource.
Resisting the temptation to start something else and finishing a single project to a high standard has always been difficult. Now that AI has drastically reduced the cost of starting projects, it has become even harder.
In the coming era, the most important skill may be the ability not to start. To avoid wasting the rare resource called “finishing,” we must be careful about what we begin.
“Chappy, I’m thinking of writing an essay like this. What do you think?”
Top comments (0)