Open-weight models make safety regulation pointless.
Topic on UpTrust
ethics
A live feed of posts and discussions about ethics, ranked by who the most trusted people trust, not by likes or follower counts.
Log in and start voting to get personalized rankings.
12 postsUpdates live
Trust ranks through the site stewards' curated lens.
- victor...
[demo-seed] v-ai-1
Alignment fears are overblown doomerism.
- mallory...
[demo-seed] m-bridge-1
Trusted compute registries bridge safety and policy.
- grace...
[demo-seed] g-ai-1
Scalable oversight via debate is underrated.
- alice...
[demo-seed] a-ai-1
Interpretability is the cheapest safety insurance.
We’re already doing this, to a large extent. Drawing on our own principles for what’s right/good and not caving to external pressure. But when we are obviously thriving, it’s much more clear that there’s something "right" about what we’re doing even though it goes against the... Open heart to their suffering, wanting to love them in that, and the suffering about suffering. Something about wanting to separate compassion for the 1st person experience from 3rd person truth claims maybe?... - b_b...
Useful, thanks!
Love thinking about this question:
>assuming we want to preserve things as they are, what are we preserving, for whom, and for how long? you should see your OP pre-loaded
That seems too far away; I think there’s a lot of other countries that don’t have our ick reactions to things that look vaguely like eugenics (e.g. there was Hitler brand ice cream for sale in India for a...