What I think of AI

I figured I’d take some time to jot what I think about AI. I’ve never formulated a cohesive opinion on it and this should help me whenever I have to discuss it or evaluate its part in a project. Let’s break this whole thing down. I will likely add more in time, but this is a good start for now. My AI journey I started using CoPilot and ChatGPT very early when I first heard about them (2022-3ish) and realised they could be a very powerful programming tool. I initially found them to be like an extremely powerful autocomplete and very helpful for generating smaller chunks of code. Things I still consider them to be good at. I experimented with using ChatGPT to create a Dungeons and Dragons DM after my groups moved away. It couldn’t even come close. I used ChatGPT in a hackathon to communicate Grasshopper scripts to it ^[1], it did very well. This project has been continuously cited by various academic papers which has been quite exciting. Eventually I realised I might be leaning on AI more than I’d like and decided to stop using it as an experiment. I’ve still never shipped code AI generated that I haven’t read, and I’m glad in retrospect I at least had that pride in my work. I’ve largely stopped using AI for programming (occasionally using it for debugging). I’m at an impasse where I cannot decide on the correct point in a programmers career to hand them AI. As an Architect my first year I was told to never use CAD and only learn using pens and pencils to learn the importance and intent of every line. I think AI should be treated the same way, but I don’t know how many years or which concepts I should have mastered first. ...

January 21, 2026 · clicky