Conversation

This Tweet was deleted by the Tweet author. Learn more
It sounds like you've been talking to some rigorous and honest people, wise enough to know when a solution is terrible and won't really help. This seems good. If at some point you start to feel like you're being talked into believing in some solution, ping me to unconvince you.
4
161
2. If you heavily modify your brain to 'merge with AGI', your modifications may change your values and motivations in harmful ways. There's an alignment problem between humans and their sufficiently modified descendants, not just between humans and de novo AI.
1
10
(In some ways a harder one, since we can in principle design AI for understandability, whereas human brains grandfather in a lot of important-to-preserve complexity.)
1
7
Show replies