Continued from conversations during the AI for Epistemics Hackathon, plus during AI4E workshop
Goal of fieldbuilding: Find great people
- Specific founders to work on projects
- Founders are super important
- How do you nerdsnipe them into working on AI4E?
- Influencers like Andy Matuschak, Scott Alexander
- Scott’s writing on prediction markets kind of precipitated Manifold
- Just need someone to sit down and make it their “thing”?
- Funders?
- Interestingly, less of a problem compared to other nascent fields. Between Ought the nonprofit, and OpenPhil dollars, FLF interest, and general interest in the space, seems like we could access money; we just don’t have shovel-ready uses for it.
- alternative essay title: “how to turn money into a new field”
Organize one-off events
- Workshops
- Conferences
- Hackathons
- Hackathons draw in talent; are a mini-competition to sort people & projects; act as a gym, a place for people to stretch their “building” muscles
- Productionizing the work out of hackathons?
- Eg getting more deployment of Panda & Charlie’s research
- Hackathons for the LLM era:
- Promptathon
- Evaluathon
- Model Trainathon
- Seems like hackathons focus on “gluing together interesting datasets & UI”, which historically is okay for things software APIs.
- (but also, greenfield hackathons don’t necessarily have great track records of getting new projects and startups actually shipped)
- But there’s a different kind of work which is about improving quality, “make evals go up
Pop-up cities, multi-org retreats
- Bring together people already working in the field, to continue their fulltime work on it, for a few weeks
- Eg EA Bahamas — this got Austin into the EA community, actually met people & made friends (vs something like a conference)
- Jay: Harder for people with kids to attend though
- Run a team residence at FAR — bringing teams in. Community, work on your normal stuff
Fellowships, Residencies
- Structure: bring in people without specific projects; give them some mentorship, support, funding to do work
- Typically more common for research & writing; less common for products
- Upskilling people vs producing concrete outputs
- Have the talent built up for a crisis
- Fellowship programs that seem good:
- Residency at labs (eg Anthropic) seems good