There's really no need to move at the speed of AI

Nowadays, I can ask for something to be built, even without properly qualifying its need, and BAM(!) there is a version of it waiting for me. That is an incredible thing, especially as someone who remembers how much work used to go into getting even a rough idea onto a screen. Something being easy, doesn’t quite make it valuable though. When there is always an opportunity to do more, how do you decide you have done enough?

Over the years when building software, there was always a fairly useful limit on our enthusiasm: eventually, someone had to do the work! You can have all the ideas you want, but implementing them forces you to make choices about what deserves your time. AI has flipped this on its head. Something you would have put aside because it needed a few days (of consideration) can now be worth trying immediately. I enjoy that freedom, although I think it also makes it easier to skip a decision without noticing. We go from “would this be useful?” to “let’s see what it looks like” and before long we are discussing how to improve something we haven’t yet found a good reason to keep.

As developers we are particularly susceptible to this because making things is part of what we enjoy. Give us an interesting problem and we can happily spend hours on it. There is satisfaction in seeing something work, and with AI we can get to that moment in a few mins of prompting. The difficulty is that a working feature answers only some of the questions about whether it belongs in a product. Someone still has to use it, understand it and find it useful enough to come back. Those things take time, sometimes an inconvenient amount of time when you already have five ideas for what to build next.

Years ago, I wrote about learning to do less whilst pair programming. One thing that stayed with me was how deliberately my colleague kept his work environment focused on the thing he was doing. I thought that was genius. It seems even more relevant now, when the tools can help us start so much work at once. We have become very good at removing the difficulty of starting; I wonder how much attention we are giving to the discipline of choosing.

This is where “don’t move at the speed of AI” starts to mean something to me. I want to be able to benefit from these tools and still decide what a worthwhile day looks like for myself. If a task takes less time than it used to, surely I should have some say in what happens to the time left over? I could use it to understand the result properly, talk to someone who uses the product, or simply finish for the day. Yet it is quite easy to treat that space as an invitation to fit in another task.

I think being able to leave some of that capacity unused is going to take practice. There will always be another improvement available, and some of them will be very good ideas. I still need to be able to say, this is enough for today, and mean it.