I’m a generalist and open sourcerer that does a bit of everything, but perhaps nothing particularly well. I’m also the Co-Director of Kairos, an AI safety fieldbuilding org.
I was previously the AI Safety Group Support Lead at CEA and a Software Engineer in the Worldview Investigations Team at Rethink Priorities.
SB 1047 is a critical piece of legislation for AI safety, but there haven’t been great ways of getting up to speed, especially since the bill has been amended several times. Since the bill’s now finalized, better resources exist to catch up. Here’s a few:
A four-page summary of the bill [written by bill proponents]
A recent post from Zvi Mowshowitz explaining the latest version of the bill
A summary of the latest round of amendments [written by bill proponents]
If you are working in AI safety or AI policy, I think understanding this bill is pretty important. Hopefully this helps.
I’m so excited about the risk-based portfolio builder!
This is one of the most exciting forecasting proposals I’ve read in some time. I’m thoroughly impressed, and I’m really eager to see more work building on top of this.
If someone ends up running an algorithmic forecasting tournament like the ones proposed in this article, count me in ;)
I’ll note that I think that choosing to prioritize resources (or support) for top universities doesn’t imply that one thinks that EAs need to be any less altruistic or truth-seeking. One can prioritize groups at top universities while (for example) maintaining the same threshold for what a good organizer or group member looks like. These kinds of strategies might result in less diversity overall, but I don’t think they result in having less altruistic people.
Could you clarify what you mean by margin-based reasoning in this context?
I agree that it’s important to consider both needs and interests. Ultimately, a branding strategy should be embedded in a larger theory of change and strategy for your group, and that should determine which audiences you reach out to.
Regarding the latter, I agree that an interest in, say, hacker culture, does not adequately describe all people interested in CS. It might actually leave out a bunch of people that you should care about joining our group. At the same time, branding is all about tradeoffs, and you have to pick which things you cater to. Spread too thin, and you risk making the content too unappealing.
It’s hard to give empirical data on this because I don’t think we have a good track record of actually collecting it. I would be curious about groups trying things like A/B tests to refine their strategies.
So yeah, most of this is backed by a mix of anecdotes from organizers, marketing know-how, and some of what I learned running my old AI Safety Initiative. This is why I want to emphasize that most of this is to be taken as provisional rather than the final word on the matter.
A few years ago, I started an EA university group in Chile, got funding and support through UGAP, then got advising through 80k and continued mentoring through OSP. This year, I got hired at CEA’s Groups team.
It’s difficult to trace counterfactual impact, but I suspect this wouldn’t have been possible if UGAP hadn’t helped me get a group started, or if CEA wasn’t able to recognize good talent outside top universities.
I’m biased now, but I think people miss how most of our team’s efforts go to scalable support, which doesn’t necessarily target top universities (and includes universities like mine). Selecting top universities is a good heuristic for getting high-quality group members, but you also get diminishing returns fast because you’re bottlenecked by having great organizers at these universities who are in need of support. This, alongside the fact that there is such a thing as great groups outside of top universities, is why we still spend a lot of time, resources, and money on other universities worldwide.
I think this is fine: Epoch’s work appeals to a broad audience, and Nat Friedman is a well-respected technologist.
I read your post while I was writing up the wiki article on Shapley values and thought it was really useful. Thanks for making that post!
Quick poll [✅ / ❌]: Do you feel like you don’t have a good grasp of Shapley values, despite wanting to?
(Context for after voting: I’m trying to figure out if more explainers of this would be helpful. I still feel confused about some of its implications, despite having spent significant time trying to understand it)
There’s a lot in this post that I strongly relate to. I also recently left CEA, although after having worked for a much smaller period of time: only 6 months. To give some perspective on how much I agree with Lizka, I’ll quote from the farewell letter I wrote to the team:
So I want to second Lizka’s thoughts: I feel very honored to have worked with them.