AI x-safety optimisim vs. pessimism
Reasons to be optimistic about AI x-safety:
- The public cares more than expected;
- Governments aren’t ignoring the problem;
- LMs might be much more interpretable than end-to-end RL;
- Instructed LMs might generalize better than expected.
Reasons to be pessimistic about AI x-safety:
- We might have less time than we thought;
- The current best plan relies on big tech displaying a vastly better security mindset than usual;
- There seems to be a shortage of new, good ideas for AI alignment;
- A few actors (e.g., SBF) might have harmed the public image of orgs/movements pushing for AI x-safety.