Post by @Federation_Bot

Re: https://fedi.copyleft.org/@ossguy/116422078885364618

I'm a systems admin. When I deploy and maintain things, I need to know that I can trust those things and the people who develop them. AI generated slop code isn't just about development - people who act like it is are missing out on a whole group of stakeholders. So by saying "Yeah we're cool with AI; bring it on in!" you're taking an entire group of people who have voices and needs of our own and telling us to piss up a rope - accept the AI generated security holes and instability and like it, plebs.

And I won't do that. That shows me plainly that I can't trust the people and the project that allows it. So I have to walk away and find something else.

The people I serve as their sysadmin deserve better than just giving up and rolling over while AI generated crap floods the market. Yes, it makes my life infinitely harder because I am constantly playing "AI-Whack-a-Mole" with various pieces of software and trying to juggle available features with security and privacy, but the people who trust me do so precisely BECAUSE I am willing to put in the work. They know that I will absolutely, without hesitation, go to war against these AI grifters and supporters who want us to just roll over and take it. And I'm not going to stop just because "well some people might lie!" - That's not a reason to stop; that's an excuse for boosters to do what they were going to anyway.

AI generated slop isn't just a developer issue. It never has been. And I am sick and tired of pretending that it is.

Denver Gingerich

@ossguy@fedi.copyleft.org 5 mo. ago

@davidgerard @wwahammy @silverwizard @firefly_lightning @cwebber Yes, which is why it's important to allow people to identify when they have used LLM/AI assistants to help. New contributors will see this is the norm, and then it will be easier to help them, because we'll know a bit about where any potential knowledge gaps might be coming from.

If we "ban" LLM/AI-assisted contributions, people will use them anyway but hide their use, which is a trickier problem to solve.

5 Replies

2 Boosts

David Gerard

@davidgerard@circumstances.run 5 mo. ago

@arthfach yeah. one of my frequent tasks as a sysadmin was to assess some lump of code I'd first heard of a day ago.

Vibe code would be unlikely to pass the vibe check.

At worst I'd have to treat it a bit like I did Magento, i.e. a concrete sarcophagus and a 30km exclusion zone.

ShnizmuffiN

@shnizmuffin@toots.inbutts.lol 5 mo. ago

@davidgerard fuck Magento!

Arthfach 🐻

@arthfach@social.arthfach.com 5 mo. ago

@davidgerard That's exactly the case for me as well. "Hey here's a thing we need / I found; it's that 'open source' you like. Take a look and tell me about it by tomorrow morning."

Vibe code has never left me with feelings of anything except concern. It just shows so much carelessness.

David Gerard

@davidgerard@circumstances.run 5 mo. ago

@arthfach more often it was "we ABSOLUTELY REQUIRE this proprietary junk/version 0.0.x library/thing that is known not to work, it's a business imperative", but we knew how to handle that sort of thing ;-)

in the case of a lump of vibe code I would likely (a) write a diplomatic memo warning them in writing of the likely consequences we can accept no responsibility for (b) use the Chernobyl isolation method (c) avoid the implementation as long as possible. A bit like requests for GDPR disasters.

David Gerard

@davidgerard@circumstances.run 5 mo. ago

@arthfach that said, our CTO was all in on using GitHub Copilot for everything, and the worst coder in the office thought it was the best thing EVER and churned out features at lightning pace! Each causing three more bugs.

1 more reply

bonfire.cafe