‘built for user trust’
https://www.youtube.com/watch?v=y2m3zjpTOcY&list=UU9rJrMVgcXTfa8xuMnbhAEA - video
https://pivottoai.libsyn.com/20251203-googles-antigravity-ai-vibe-coder-wipes-your-hard-disk - podcast
time: 5 min 57 sec
deleted by creator
trusting
an AIglorified autocomplete to do anything with your file system was your first mistakedeleted by creator
Lol, honestly, that’s on you for having incorrectly adjusted priors. Whenever I catch wind of a prompting snafu, I immediately assume I’ll hear the rejoinder of “You must be prompting it wrong”.
“how foolish of you to believe the 24/7 barrage of propaganda about AI agents”
The hardest you can nonconsensually destroy a D wile still being allowed on pornhub.
To (somewhat) reiterate a potentially extreme position of mine, shit like this is why society needs to actively avoid adopting new software for the time being, if not actively cut back on software usage whenever possible.
Whilst the IT industry was already a failure-ridden mess which actively refuses to learn before AI was a thing, the rise of AI and “vibe coding” has made things so much worse in practically every regard. At this point, any new software should be treated as a liability until proven otherwise.
while I relate to this take and it could even work out for offline systems, online systems need constant updates to protect yourself against software vulnerabilities. i don’t think we’ve reached the point where FOSS software is so infiltrated that a Trojan would be less malicious than slop code
That is a pretty solid point - constant updates are something online systems need on a regular basis.
AFAIK FOSS has avoided being slopified by AI, so we should hopefully be pretty far from that point
Promptfondlers are trying mightily to slopify FOSS whether it likes it or not.
certain projects like Fedora and Bitwarden are considering it, so I’m not too pleased about that
