pull down to refresh

frontier model companies are evil ... but I trust the slop code they generate without reading it because I asked the model if it has any bugs

Sheesh starting to sound like a AI bear

reply
82 sats \ 4 replies \ @k00b 19 Jun

I'm not. I think people are being a bit reckless though while shaming others for not being reckless enough.

reply

Yeah @optimism crushed my dreams this I stopped messing with vibe coding

reply
225 sats \ 2 replies \ @k00b 19 Jun

The threat levels are different for everyone/thing. If you're making stuff for yourself that's noncritical and on an isolated/firewalled computer, being careful with dependencies, you are probably fine. If you're making static pages/games for other people to use, while being conscious of your own security, I'd say that's okay too. When you start dealing with yours or other people's money and privacy, it only takes one small mistake to jeopardize that and LLMs are far from perfect. Best use for LLMs in money/privacy domains is not yolo vibing. The best use is making the code higher quality, more defensive, and more deeply scrutinized.

reply
146 sats \ 1 reply \ @k00b 19 Jun

I think that's one reason why we see people pivoting to AI. Working on low threat level tech lets you vibe harder.

reply

Yeah I agree. Plus the time that is needed.

reply

welcome to my week of reviewing downstream npm packages

reply