🧩 Philosophy 1d ago · beren

Three Phases of Capabilities and Alignment

Less Wrong
View Channel →
Source ↗ 👁 1 💬 0
Cross-posted from my personal blogEpistemic Status: Reality is messier than the elegant procession of 'phases' presented here but, nevertheless, I do think this gets at an important way to think about how alignment and capabilities are intertwined and how the alignment challenge becomes increasingly abstract as the underlying autonomy and generality of the models improves over time. A thought struck me recently which I thought was interesting. As AI has progressed through different technological

Comments (0)

Sign in to join the discussion

More Like This

📰
Porphyry
Stanford Encyclopedia of Philosophy · 11h ago
Measure Optimal AI Share of Labor Costs Instead of AI or Human-Only Performance
LessWrong · 15h ago
Misaligned Models Still Recognise When Their Responses Are Harmful. Can Training Make Them Care?
LessWrong · 16h ago
Lightcone Commons: what I funded and why
LessWrong · 17h ago
📰
Live disagreements with people around me
LessWrong · 17h ago
📰
OpenAtom and Elemental
LessWrong · 17h ago