- mechanistic agents vs non-mechanistic one
- enforcing ethics on AI models; need moral philophers
- how do you instantiate a moral philosopy in an AI?
- How do we make a virtuous AI? Can we?
Dario Amodei: We Don’t Know if the Models Are Conscious
- diminishing returns of intelligence? – marginal productivity of intelligence, like marginal productivity of land and labor
- maybe for real-world problems having 100 million genius level ai’s rather than ten “god” ai’s is more effective. Some problem, like go or chess require extremely high levels of intelligence, but perhaps most real-world problems are not like that.