lukeprog's Quick takes

Therefore, we think AI policy work that engages conservative audiences is especially urgent and neglected, and we regularly recommend right-of-center funding opportunities in this category to several funders.

Should the reader infer anything from the absence of a reference to GV here? The comment thread that came to mind when reading this response was significantly about GV (although there was some conflation of OP and GV within it). So if OP felt it could recommend US "right-of-center"^[1] policy work to GV, I would be somewhat surprised that this well-written post didn't say that.

Conditional on GV actually being closed to right-of-center policy work, I express no criticism of that decision here. It's generally not cool to criticize donors for declining to donate to stuff that is in tension or conflict with their values, and it seems that would be the case. However, where the funder is as critical to an ecosystem as GV is here, I think fairly high transparency about the unwillingness to fund a particular niche is necessary to allow the ecosystem to adjust. For example, learning that GV is closed to a niche area that John Doe finds important could switch John from object-level work to earning to give. And people considering moving to object-level work need to clearly understand if the 800-pound gorilla funder will be closed to them.

1. ^{^}

   I place this in quotes because the term is ambiguous.

I think it might be a good idea to taboo the phrase "OP is funding X" (at least when talking about present day Open Phil). 

Historically, OP would have used the phrase "OP is funding X" to mean "referred a grant to X to GV" (which approximately was never rejected). One was also able to roughly assume that if OP decides to not recommend a grant to GV, that most OP staff do not think that grant would be more cost-effective than other grants referred to GV (and as such, the word people used to describe OP not referring a grant to GV was "rejecting X" or "defunding X"). 

Of course, now that the relationship between OP and GV has substantially changed, and the trust has broken down somewhat, the term "OP is funding X" is confusing (including IMO in your comment, where in your last few bullet points you talk about "OP has given far more to global health than AI" when I think to not confuse people here, it would be good to say "OP has recommended far more grants to global health", since OP itself has not actually given away any money directly, and in the rest of your comment you use "recommend").

I think the key thing for people to understand is why it no longer makes sense to talk about "OP funding X", and where it makes sense to model OP grant-referrals to GV as still closely matching OPs internal cost-effectiveness estimates.^[1] 

For organizations and funders trying to orient towards the funding ecosystem, the most important thing is understanding what GV is likely to fund on behalf of an OP recommendation. So when people talk about "OP funding X" or "OP not funding X" that is what they usually refer to (and that is also again how OP has historically used those words, and how you have used those words in your comment). I expect this usage to change over time, but it will take a while (and would ask for you to be gracious and charitable when trying to understand what people mean when they conflate OP and GV in discussions).^[2]

Now having gotten that clarification out of the way, my guess is most of the critiques that you have seen about OP funding are basically accurate when seen through this lens (though I don't know what critiques you are referring to, since you aren't being specific). As an example, as Jason says in another comment, it does look like GV has a very limited appetite for grants to right-of-center organizations, and since (as you say yourself) the external funders reject the majority of grants you refer to them, this de-facto leads to a large reduction of funding, and a large negative incentive for founders and organizations who are considering working more with the political right.

I think your comment is useful, and helps people understand some of how OP is trying to counteract the ways GV's withdrawal from many crucial funding areas has affected things, which I am glad about. I do also think your comment has far too much of the vibe of "nothing has changed in the last year" and "you shouldn't worry too much about which areas GV wants or want to not fund". De-facto GV was and is likely to continue to be 95%+ of the giving that OP is influencing, and the dynamics between OP and non-GV funders are drastically different than the dynamics historically between OP and GV.

I think a better intutition pump for people trying to understand the funding ecosystem would be a comment that is scope-sensitive in the relevant ways. I think it would start with saying:

Yes, over the last 1-2 years our relationship to GV has changed, and I think it no longer really makes sense to think about OP 'funding X'. These days, especially in the catastrophic risk space, it makes more sense to think of OP as a middleman between grantees and other foundations and large donors. This is a large shift, and I think understanding how that shift has changed funding allocation is of crucial importance when trying to predict which projects in this space are underfunded, and what new projects might be able to get funding.

95%+ of recommendations we make are to GV. When GV does not want to fund something, it is up to a relatively loose set of external funders we have weaker relationships with to make the grant, and will hinge on whether those external funders have appetite for that kind of grant, which depends heavily on their more idiosyncratic interests and preferences. Most grants that we do not refer to GV, but would like to see funded, do not ultimately get funded by other funders.^[3]

[Add the rest of your comment, ideally explaining how GV might differ from OP here^[4]]

1. ^{^}

   And another dimension to track is "where OPs cost-effectiveness estimates are likely to be wrong". I think due to the tricky nature of the OP/GV relationship, I expect OP to systematically be worse at making accurate cost-effectiveness estimates where GV has strong reputation-adjacent opinions, because of course it is of crucial importance for OP to stay "in-sync" with GV, and repeated prolonged disagreements are the kind of thing that tend to cause people and organizations to get out of sync.

2. ^{^}

   Of course, people might also care about the opinions of OP staff, as people who have been thinking about grantmaking for a long time, but my sense is that in as much as those opinions do not translate into funding, that is of lesser importance when trying to identify neglected niches and funding approaches (but still important).

3. ^{^}

   I don't know how true this is and of course you should write what seems true to you here. I currently think this is true, but also "60% of grants referred get made" would not be that surprising. And also of course this is a two-sided game where OP will take into account whether there are any funders even before deciding whether to evaluate a grant at all, and so the ground truth here is kind of tricky to establish.

4. ^{^}

   For example, you say that OP is happy to work with people who are highly critical of OP. That does seem true! However, my honest best guess is that it's much less true of GV, and being publicly critical of GV and Dustin is the kind of thing that could very much influence whether OP ends up successfully referring a grant to GV, and to some degree being critical of OP also makes receiving funding from GV less likely, though much less so. That is of crucial importance to know for people when trying to decide how open and transparent to be about their opinions.

I hope in the future there will be multiple GV-scale funders for AI GCR work, with different strengths, strategies, and comparative advantages

(Fwiw, the Metaculus crowd prediction on the question ‘Will there be another donor on the scale of 2020 Good Ventures in the Effective Altruist space in 2026?’ currently sits at 43%.)

[1] Several of our grantees regularly criticize leading AI companies in their official communications [2] organizations we've directed funding to regularly propose or advocate policies that ~all frontier AI companies seem to oppose

Could you give examples of these?

Most concrete progress on worst-case AI risks — e.g. arguably the AISIs network, the draft GPAI code of practice for the EU AI Act, company RSPs, the chip and SME export controls, or some lines of technical safety work

My best guess (though very much not a confident guess) is the aggregate of these efforts are net-negative, and I think that is correlated with that work having happened in backrooms, often in context where people were unable to talk about their honest motivations. It sure is really hard to tell, but I really want people to consider the hypothesis that a bunch of these behind-the-scenes policy efforts have been backfiring, especially ex-post with a more republican administration.

The chip and SME export controls seem to currently be one of the drivers of the escalating U.S. and China arms race, the RSPs are I think largely ineffectual and have delayed the speed at which we could get regulation that is not reliant on lab supervision, and the overall EU AI act seems very bad, though I think the effect of the marginal help with drafting is of course much harder to estimate. 

Missing from this list: The executive order, which I think has retrospectively revealed itself as being a major driver of polarization of AI-risk concerns, by strongly conflating near-term risk with extinction risks. It did also do a lot of great stuff, though my best guess is we'll overall regret it (but on this I feel the least confident). 

I agree that a ton of concrete political implementation work needs to be done, but I think the people working in the space who have chosen to do that work in a way that doesn't actually engage in public discourse have made mistakes, and this has had large negative externalities. 

See also: https://www.commerce.senate.gov/services/files/55267EFF-11A8-4BD6-BE1E-61452A3C48E3

Again, really not confident here, and I agree that there is a lot of implementation work to be done that is not glorious and flashy, but I think a bunch of the ways it's been done in a kind of conspiratorial and secretive fashion has been counterproductive^[1]. 

Ultimately as you say the bottleneck for things happening is political will and buy-in that AI systems pose a serious existential risk, and I think that means a lot of implementation and backroom work is blocked and bottlenecked on that public argumentation happening. And when people try to push forward anyways, they often end up forced to conflate existential risk with highly politicized short-term issues that aren't very correlated with the actual risks, and backfire when the political winds change and people update.

1. ^{^}

   See also: https://www.lesswrong.com/posts/vFqa8DZCuhyrbSnyx/integrity-in-ai-governance-and-advocacy 

"It seems hard to predict whether superintelligence will kill everyone or not, but there's a worryingly high chance it will, and Earth isn't prepared," and seems to think the latter framing is substantially driven by concerns about what can be said "in polite company."

Funnily enough, I think this is true in the opposite direction. There is massive social pressure in EA spaces to take AI x-risk and the doomer arguments seriously. I don't think it's uncommon for someone who secretly suspects it's all a load of nonsense to diplomatically say a statement like the above, in "polite EA company".

Like you: I urge people who think AI x-risk is overblown to make their arguments loudly and repeatedly. 

I think the difference between me and Yudkowsky has less to do with social effects on our speech and more to do with differing epistemic practices, i.e. about how confident one can reasonably be about the effects of poorly understood future technologies emerging in future, poorly understood circumstances. 

This isn't expressing disagreement, but I think it's also important to consider the social effects of our speaking in line with different epistemic practices, i.e.,

• When someone says "AI will kill us all" do people understand us as expressing 100% confidence in extinction, or do they interpret it as mere hyperbole and rhetoric, and infer that what we actually mean is that AI will potentially kill us all or have other drastic effects
• When someone says "There's a high risk AI kills us all or disempowers us" do people understand this as us expressing very high confidence that it kills us all or as saying it almost certainly won't kill us all.