The best AI success

Duolingo argument from Acquired podcast.

I’m still convicted: The more useful your tool is, the easier it is for AI to make it more useful. If your tool is a pain and you use AI to fix it, that’s a recipe for disaster.

This happens in programming! Easy languages are easy for AI. Painful languages are made better with AI, but they’re still painful in proportion to their pre-AI painfulness.

kubernetes example

git example

code example: AI works well with code that is already good. Messy code is hard to edit.