extending an enormous "bruh" to the people who are making a spreadsheet of all the mainstream programming languages and tools that you "shouldn't use" because they have some tests written by Claude that probably wouldn't have been written without it (ie. it increases correctness). cmon man.

yes there are people writing slop but almost all people who give a shit and work on production systems don't (and are probably busy getting their day ruined by jackasses wielding chatbots even if they do use them themselves).

leftists gobble up purity politics better than devout Catholics.

@leftpaddotpy I'm afraid I'm mostly with them on this. I'm fully aware I'm going to need to compromise and use some of this stuff, but at the very least I want to know who is using it for what so I can make informed decisions. The mere fact that it's potentially so opaque is a huge part of the problem. My urge to set all the tech in my life on fire and go farm goats is intensifying day by day.
@leftpaddotpy (My anti-AIness isn't particularly morality or principles, I just think it's going to have rapid cognitive and quality impacts anywhere it's applied, and I want my systems to have as little exposure to that as possible.)
@leftpaddotpy my perspective has changed quite a lot since I've listened to https://securitycryptographywhatever.com/2026/03/25/ai-bug-finding/. These are people who know a thing or two about security and you can listen to them trying to understand the consequences of LLMs on the field. Afterwards I don't feel any less annoyed by AI, just more like LLM based security analysis will be needed to defend against those abusing these tools
AI Finds Vulns You Can’t With Nicholas Carlini

Returning champion Nicholas Carlini comes back to talk about using Claude for vulnerability research, and the current vulnpocalypse. It’s all very high-brow ...

@leftpaddotpy even some of the smallest models (that I can run on my own computer!) are extremely good refactor engines in comparison to like... AST manipulation or a bunch of futzing with regexes

@traumaphoenix i have been caught doing `some-linter | claude --model haiku 'fix these lints plz'`

of course i would rather the linter be able to fix them but i am overcommitted on tooling fixing in general as it is