AI code comments are getting unbearable #
AI’s have appeared to learn a new trick to help them with context. It looks like this was done to help an agent returning to a code base for improvements. What they do is to add a lot of verbose comments on why something does the way it is. An agent reading it gets better context instead of inferring the motivations behind the code. Unfortunately they don’t always do a proper job, they leave comments which are “intent” but not done. Like a “we did X here because …” but actually no X was done. Agents returning to the code base come across this and are angry. God knows why the other agent left a comment like that.
I am going to write a lint rule to stop writing comments and see how it goes.
source & further reading
pradeeproark.com — original article