You know how I can tell AI coding is getting more popular with software companies?

Updates are coming quicker, and they’re breaking more things. New features are being rolled out at a rapid rate but they’re broken or so ill-conceived it’s clear they went “from idea to implementation in seconds.”

I was thinking about that triangle problem where you can only have two out of three of the things.

Here it’s:
Faster, more productive output
Stable, useable software
Lower overhead

Companies, in search of more profit to gift to their board and C Suite are choosing 1+3 but they should be choosing 1+2

I’m not opposed to AI coding at all by the way. I’m just opposed to the naive capitalist narrative that just firing half the devs and replacing them with AI will create the best outcomes. At the very least you’d need more code reviewers, testers, and bug finders.

A doodler and computerer. I like permissive/permissionless open source, smart contracts, p2p systems, room-scale VR, and NixOS.

shazow.net

Currently: WhatsABI

My theory is that code mostly completed by LLMs is being entirely reviewed by LLMs, and in many cases, even the bugs are being fixed by LLMs without much guidance (besides the bug report)

I’m implying that they are rushing out bad software because they cut their teams dramatically. Huge bugs in the entire Google suite, slack, spark mail, adobe suite. It’s like everything I use all day long is just constantly broken for the past few months in a way that it has never been before. Tried access multiple computers and asked others about this and they’re all experiencing it

Are you implying this is because they're using AI or because they're rushing to service people who are using AI? (Move fast and break things etc)