Well, I think its official: we are clearly living through a moment when the ability to make things is no longer scarce.
So the important ability now is deciding exactly what deserves to be built.
There’s something I’ve always liked to say about this. On my birthdays, for example, I like doing things that I actually enjoy doing. Like, I literally don’t do tasks. I don’t do homework on my birthday. I don’t do things I don’t want to do.
And what do I want to do?
Sometimes I design interfaces. I’ll make a few pages for something I had in mind, create some animations, play around with tiny details. Little things that ultimately change no outcome, produce no obvious result in my life, and probably won’t go anywhere.
And that’s fine.
I mean, yes, you can code for fun too, like we used to 👀. But there is also an instrumental value to almost everything we do. Most of the time, actually almost every day, you do things you don’t necessarily feel like doing because, through doing them, you’re trying to get some kind of result that you do want.
And once you care about the outcome, the question changes:
A lot of the outcomes we want don’t necessarily come from programming. And even within the things we can build, not every improvement is equally valuable. You can spend hours making something faster, prettier, cleaner, more polished, more technically impressive and still not change the outcome enough.
And you can see this even more clearly if you go to the other extreme 👀
There’s something I find really interesting about AI-generated image ads or AI-written text. In both cases, yes, most of the times it looks bad or reads horribly, and yes, it could probably have been written much, much better, but the message doesn’t disappear.
And if a message comes to you through a genuinely bad website but manages to grab your attention, it still has value.
Design and all those tiny details are very concentrated in a very small number of apps where they actually matter. Social networks, for example, eventually need incredibly extensive systems for understanding how their users behave (yes, this is a reference).
But back to the point, that’s also the problem: you can do many things, but you can only put serious effort into very, very few of them.
You can do everything, but you cannot do every single thing they can do
When I hit the tokenmaxxing phase (using AI to code at midnight), much earlier than I’d like to admit, one of the first websites I built was SoundAir, which I posted here a few years ago.
The problem with SoundAir was that it was horrendously bad performance-wise 😭 (it was built with ChatGPT), and the idea itself was also bad. So it was bad twice.
But I think that was the first time I felt that enormous rush of dopamine from realizing: wow, I can make things with artificial intelligence.
And from that moment on, it just started going down… I simply got used to saying: okay, I can do everything. At most, I’ll learn to prompt all of this, and even that will be easier with the technical background I already have.
Absolutely everything is possible.
And if I run into some technical challenge, it will most likely be a temporary technical challenge, because eventually it will be possible to solve it.
That’s just how it is.
So we know empirically that artificial intelligence can massively increase what a person is capable of doing. But one way not to become overwhelmed by that is to stop trying to build everything.
Decision overload is basically that: having more options obviously clouds things a little. It clouds our sense of what we should actually be doing.
If before we were constrained mostly by our capacity for implementation, today we have basically arrived at the constraint of attention.
Attention and, eventually, direction: the direction that each human being can give to a product, or to anything else.
I’ve tried racing against time just to see where the limits are. And yes, I think I’ve finally reached… I think somewhere between this year and last year, the point where I understand exactly what my capabilities are and what kinds of things I’m capable of accomplishing.
So the ultimate promise of artificial intelligence is that we can do more. But eventually you have to ask, in the broadest possible sense: was that actually what we needed?
Because a huge part of what we want to accomplish in life has very little to do with the software itself.
It had to do, from the very beginning! with the scarcity of attention.
Basically, with breaking through the growth barrier: talking to people, connecting with users. All of that requires effort that artificial intelligence doesn’t postpone for you, and that eventually becomes part of your own strength.
The worst possible combination would be to code your website by hand and then use artificial intelligence to contact all your users through automated messages.
That’s basically the worst possible combination.
There are points where AI gets in the way. Where it simply cannot do the thing. And even if it technically could, because ultimately it’s a model capable of spitting out words and you can give it instructions so it spits those words out in the correct format and the correct order, sometimes getting it (the prompting part) there takes so long that it simply isn’t worth it in terms of speed.
An example of that: I was choosing Expo for my very first apps, and only with AI did I make the switch to SwiftUI. Yes, SwiftUI is much better, native, etc., but I need the app today, not when it’s possible.
And yes, maybe in the future we’ll eventually manage to imitate human behavior in writing.
Actually, we will.
I mean, when Stack Overflow banned AI-generated answers for code, I protested because I said: hey, it will get better. It doesn’t make sense to do this.
And I think the same thing is going to happen with writing.
Eventually, this text that I’m writing by hand will be something an AI can write in a style so specifically mine that it becomes indistinguishable from something I would actually say.
And it will happen.
But human connection is incredibly broader, incredibly harder to overcome, and therefore much, much harder to replace. And it’s something very few people seem willing to really dig into. You can’t automate that.
So if you actually want to create something meaningful, don’t create just for the sake of creating.
Even when creating is cheap.
Even when you can turn an idea into a landing page in twenty minutes, build the product in a weekend, and have something working before you’ve even had enough time to decide whether it should exist.
Because eventually, you’ll get tired.
You can prompt a thousand projects. You can keep building and building and building because suddenly the cost of doing so feels almost nonexistent.
But your attention is not infinite.
And eventually, you’re going to exhaust yourself building things that were never going anywhere in the first place.
That’s the strange thing about this new abundance. The fact that something has become incredibly cheap to create doesn’t suddenly make it valuable to create.
There are still things that depend on the world. Things that have nothing to do with creating a landing page and calling it a day, but with talking to people, reaching someone’s consciousness, understanding what they care about, and building something that actually means something to them. ❤️
And you cannot simply prompt your way through all of that.
So look at what you’re spending this new capacity on.
If it’s for learning, it’s incredibly good. If it’s for learning and creating, that might be the best combination in the world.