Artificial intelligence doesn't get tired, and that is the starting point for everything I think about how to use it. Tokens are cheap, energy is cheap, and almost all the tech capital in the world is sitting right there, so if I try to keep up with its pace it will run me over, because a race against something that never rests can't be won by wanting it more.
That is why I think we need a limit, both on the time I spend with it and, above all, on the energy I give it.
It doesn't get tired. I do.
A person works for a while and then performance drops, because we need to eat, sleep and disconnect, whereas an AI answers the same at one in the afternoon as at three in the morning and always has another version ready, another idea, another option to compare.
That is the trap: since it can keep going, you feel you should keep going too, so you ask for one more variation, try one more thing, and by the time you notice four hours have gone by and your head is fried even though the physical effort was close to zero.
Mental fatigue warns you far less clearly than physical fatigue does, and by the time you notice it you have already made several bad decisions without realizing.
First, pencil and paper
The first thing I do to learn how to use it is write down on paper, with a pencil and no screen, what I want to achieve. It sounds like a small thing, but if I can't explain in a few lines what I'm after, I'm not ready to ask a machine for anything, and since it always answers, it will hand me something anyway, something plausible that resembles what I wanted without quite being it.
Writing first forces me to define three things:
- What I want to achieve, said in my own words and in a way I can reread the next day and still understand.
- What "done" looks like, so I know clearly when I'm allowed to stop.
- How much time and energy I'm willing to give it, decided before I start and not halfway through, when I'm already invested.
That gives me an edge, and without an edge the project expands on its own until it fills all the space I leave it.
Every project is a debt
Everything I start costs me energy, and that cost reaches forward in time, because each new project is a cognitive work debt: something I have to understand, review, maintain and carry in my head, and which represents life time I stop spending on other things.
Before starting, I ask myself what I'm going to stop doing because of this, whether it's time with my family, a book, sleep or some other project that mattered more and is now waiting. The question weighs more right now precisely because everything looks easy, and when starting costs nothing it becomes very simple to start too many things at once.
Code is the easy part
It's true that AI writes code fast and that it often works, but a project includes a lot more than code. Take a small app to organize something: the AI builds it in an hour, and then comes everything else, which takes up most of the work:
- Deciding what gets built and what stays out, a task no tool can settle for you.
- Testing it with real people, who use things in ways you never imagined.
- Fixing, fixing again and testing again until it behaves the way it should.
- Maintaining it, supporting it and explaining it to whoever comes after and has no idea how it works.
- All the back and forth that nobody mentions when they show the final result.
Reviewing what someone else did, even when that someone is a machine, is tiring too, and reviewing badly gets expensive later, when the mistake is already buried in something other people rely on. The first version coming out in an afternoon doesn't mean the project costs an afternoon, since that version marks only the beginning and I pay for the rest with my attention, which is measured in hours of my day and not in tokens.
My recommendation
My recommendation is simple, and I follow it in this order:
- Before opening the AI, I write on paper what I want and the point at which I'll call the work finished.
- I set a fixed time for the session, and when that time is up I close it, even if the conversation still has more to offer.
- I ask myself what debt I'm creating and whether it's worth what I'll stop doing to carry it.
- I stop as soon as I notice I'm tired, even if it's still ready to continue, because my best decisions come when I'm rested.
I keep using it every day, with a limit I set myself, because it will never set one for me.