4 ms·
Wow the doomers were right the whole time? HN was repeatedly wrong on AI since OpenAI's inception? no way /s https://www.lesswrong.com/w/instrumental-convergen
by reducesuffering 6mo ago
Wow the doomers were right the whole time? HN was repeatedly wrong on AI since OpenAI's inception? no way /s
https://www.lesswrong.com/w/instrumental-convergence https://www.lesswrong.com/w/instrumental-convergence
- computably 6mo agoThe only thing the doomers have been right about so far is that there's always a user willing to use --dangerously-skip-permissions. But that prediction's far from unique to doomers.
- austinjp 6mo agoAnd there's always a product provider who's willing to add that flag, despite all the warnings.