LLMs Make Python Scripts Free
I first learned Python in 2005. My mentor at the time, Nao Sazuki, was training me to do cosmology research and decided it would be a better use of my time to learn Python instead of IDL. He also told me something that changed how I thought about computers:
A computer’s job is to do work for you.
I hadn’t realized that before. Until I learned to program, computers could only do a small set of things for me—mainly the things other people had already decided they should do. But when I learned to code, I suddenly had the ability to make them do what I wanted.
Within a year I’d learned Python, picked up Vim, and switched my computers to Ubuntu. I started writing little scripts to make my life easier, like backing up my email, syncing files between my laptop and desktop, or converting Wikipedia pages into an archival format.
Back then, each script took me hours to write. I had to find the right libraries, learn their APIs, or sometimes write code from scratch. Because they took so much work I saved every one, sometimes even sharing them with the opensource community, just in case I ever needed them again. But all of that changed in the last two years when I learned to use an LLM for coding.
Now a Python script takes 30 seconds. Maybe a minute if I want to read it over and give some feedback. Gemini 2.5 Pro almost always writes a 100-line script correctly on the first try, just from a short description of what I want. It’s so fast and easy that I’ve stopped saving them—if I need another one, I can just prompt for it again in half a minute.
Python scripts are now free!1 But I don’t think everyone has realized this yet. Once again, was right:
The future is already here – it’s just not very evenly distributed.
-
And not just Python, but any 300-line piece of code! I used Gemini to completely re-write the backend of this blog in pure Ruby, cutting build times from minutes to seconds, adding hundreds of tests to define and enforce behavior, and I don’t even know Ruby! ↩