Cullen O'Keefe and I agree that people working in short-term AI policy ought to do more explicit cause prioritization. I think that some attempt at cause prioritization would be a useful resource for people in AI policy. Here are my preliminary thoughts on it:
- Priorities may look different for research, advocacy, and implementation of AI policy.
- It might be easier to evaluate the impactfulness of working on AI policy in specific domains and locales (e.g. AI fairness in U.S. health care) than more general AI policy areas (e.g. AI fairness writ large). To me, AI policy concerns in international development seem especially neglected.