There has been a lot of digital ink spilled on the usage of LLMs, but I think that their biggest advantage for me has been the decrease in activation energy to actually get to do something. I always emphasised doing things right the first time, because it’s always expensive to fix things later, but now it’s become quite cheap to have some first attempts.

I might prompt Claude for some designs of a page on the blog, for example, and I will inevitably hate all of them, but having cranked out a few in the space of an hour, I now have a very good idea what I don’t want. The activation energy was therefore minimal and I have a good idea now what I want to build myself.

This is a very significant advantage. Previously I would probably have to sit around and think, and with so many distractions around it’s much more difficult to do this now. Having the ability to prompt my way through some initial ideas, be it for code or writing, means that I don’t need to force myself to do the grunt work that much. It’s honestly very pleasant.