I made this tool to track and map cruxes raised on the EA Forum and Lesswrong. Your input ('quick rate' upvotes/downvotes, hypothesis comments, etc.) could help improve the tool. (Or let me know if it should be reshaped). 

https://uj-prioritization-prototype.netlify.app/cruxes/ 

 

Great question, thank you for working on this. An inter-cause-prio-crux that I have been wondering about is something along the lines of:

"How likely is it that a world where AI goes well for humans also goes well for other sentient beings?"

It could probably be much more precise and nuanced, but specifically, I would want to assess whether "trying to make AI go well for all sentient beings" is marginally better supported through directly related work (e.g., AIxAnimals work) or through conventional AI safety measures - the latter of which would be supported if, e.g., making AI go well for humans will inevitably or is necessary to make sure that AI goes well for all. Although if it is necessary, it would depend further on how likely AI will go well for humans and such; but I think a general assessment of AI futures that go well for humans would be a great and useful starting point for me. 

I also think various explicit estimates of how neglected exactly a (sub-)cause area is (e.g., in FTE or total funding) would greatly inform some inter-cause-prio questions I have been wondering about - assuming that explicit marginal cost-effectiveness estimates aren't really possible, this seems like the most common proxy I refer to that I am missing solid numbers on. 

I would like to see research on how to increase the welfare of soil animals and microorganisms, including on how to build capacity to do this. I think whether interventions increase or decrease welfare in expectation is determined by very uncertain dominant effects on such potential beings. I do not even know whether electrically stunning shrimp increases or decreases welfare, even if I was certain it increased the welfare of shrimps conditional on these being sentient.

This may be less fun but, for completeness, I want to present an alternative perspective -- I think I know exactly how I'd spend it and don't have any particular questions. Feel free to send the $20M over whenever works best.

Approx how much absorbency/room for more funding is there in each cause area? How many good additional opportunities are there over what is currently being funded? How steep are the diminishing returns for an additional $10m, $50m, $100m, $500m?

EA is pivoting hard into policy development, political organising, and comms work, driven from the AI side of things.

To what extent is it possible to leverage the resulting political-ability capital EA develops towards lobbying governments for the restoration of effective international aid commitments?

This might not fit the idea of a prioritization question, but it seems like there are a lot of "sure bets" in global development, where you can feel highly confident an intervention will be useful, and not that many in AI-related causes (high chance it either ends up doing nothing or being harmful), with animal welfare somewhere in between. It would be interesting to find projects in global development that look good for risk-tolerant donors, and ones in AI (and maybe animal welfare or other "longtermist" causes) that look good for less risk-tolerant donors. 

Some questions that might be cruxy and important for money allocation: 

Because there is some evidence that superforecaster aggregation might underperform in AI capabilities, how should epistemic weight be distributed between generalist forecasters, domain experts, and algorithmic prediction models? What evidence exists/can be gotten about their relative track records?

Are there better ways to do AIS CEA? What are they? 

Is there productive work to be done in inter-cause comparison among new potential cause areas (i.e. digital minds, space governance, etc)? What types of assumptions do these rely on? I ask because it seems like people typically go into these fields because "woah, those numbers are really big," but that sort of reasoning applies to lots of those fields and doesn't tell you very much about resource distribution. 

What are the reputational effects for EA (for people inside and outside the movement) going (more) all in on certain causes and then being wrong (i.e. AI is and continues to be a bubble)? Should this change how much EA should go in on things? Under what assumptions? 

Some questions which feel alive for me:

1. Should we expect risks to come from categories and places we expect to see coming, or from places we would not have anticipated afterwards? What's the proportion of black swans for the largest risks?
2. How do we incorporate a term for doing good in a way that helps us do more good in the future? Companies can sell stock, whereas nonprofits can't; the more Elon Musk does the more money he makes, even while pursuing some notion of the good, whereas the more a philanthropist gives away the less he has. This seems like a strategic disadvantage. This is more like an operational decision than like a research question though.

I'd like to see more rigorous engagement with big questions like where value comes from, what makes a good future, how we know, and how this affects cause prioritization. I think it's generally assumed "consciousness is where value comes from, so maximize it in some way." Yet some of the people who have studied consciousness most closely from a phenomenological perspective seem to not think that (e.g. zen masters, Tibetan lamas, other contemplatives, etc), let alone scale it to cosmic levels. Why? Is third person philosophical analysis alone missing something? 

The experiences of these people add up to millions of years of contemplation across thousands of years. If we accept this as a sort of "long reflection" what does that mean? If we don't, what do we envision differently and why? And are we really going to be able to do serious sustained reflection if/once we have everything we think we want within our grasp due to strong AI?

These are the kinds of things I'm currently thinking through most in my spare time and writing my thoughts up on. 

If I were committed to allocating $20M starting in 2026, the key uncertainties I would want resolved before deciding how to give fall into two clusters: questions about the near term tractability of global health interventions and questions about the risk landscape surrounding increasingly capable AI systems. What I find most neglected is research on how these domains interact.

1. How do we compare the marginal value of global health scaling vs. AI safety risk mitigation, given uncertainty around timelines and tractability?
Much of the Global Health and Development community has built strong methodologies for cost effectiveness modeling and iteration. AI safety, by contrast, remains dominated by high uncertainty and expert priors. I would want to better understand whether some AI governance or alignment work can be made more measurable such that it can be compared, even imperfectly, to global health opportunities like malaria control, vaccine delivery, or unconditional cash transfers. This feels especially relevant given @Open Philanthropy 's portfolio across both spaces.

2. What are the highest leverage AI related opportunities for improving global health outcomes?
Even before AGI, frontier systems will shape how health systems operate. We are already seeing early benefits (diagnostics, forecasting, logistics) and risks (deepfake misinformation, automated biodesign, widening inequalities). I would want rigorous analysis of whether supporting AI policy capacity in LMICs, or designing AI systems to specifically serve low resource contexts, could outperform traditional global health spending on a per dollar basis. @Evidence Action style evidence based delivery may become relevant here.

3. What is the risk profile of AI deployed in global health, and does it create new systemic vulnerabilities?
For example, how does reliance on AI created surveillance or epidemiological systems change the risk of catastrophic misuse or failure? Could global health deployments inadvertently increase existential bio risk? This feels like a question currently falling between cause areas and I am not sure any actor is systematically prioritizing it.

4. What are the optimal philanthropic strategies if AI timelines shorten significantly?
Should a donor pivot from long term global health institution building to technical alignment research or policy advocacy? Or is it more valuable to improve the resilience and welfare of populations who will otherwise be least protected from AI driven shocks? This is a core strategic question for anyone trying to maximize expected value across time.

If I had to name the most important meta question: what frameworks allow us to compare uncertain, systemic, long tail risk reduction (e.g. alignment, global governance) with concrete, short timeline health and development interventions without resorting to hand waving or relying purely on moral intuitions?

I think that expanding cross cause prioritization frameworks to include AI safety explicitly, and especially to explore AI global health interactions, is a major gap in current EA work. If I had $20M and time to wait, the research agenda I’d want to commission would sit precisely at that intersection.

What work that may reduce animal suffering looks effective (or positive at all) when we don't ignore or downplay the backfire risks via unintended indirect effects? See

• What to do about near-term cluelessness in animal welfare
• If wild animal welfare is intractable, everything is intractable
• this informal research agenda

Thanks for posting this! I will be coming back to this post next year when I'm planning debate weeks. 
 

How politicized will AI get in the next (1,2,5) and what will those trends look like?

I think we are investing as a community more in policy/advocacy/research but the value of these things might be heavily a function of the politicization/toxicity of AI. Not a strong prior but I'd assume that like OMB/Think tanks get to write a really large % of the policy for boring non electorally dominant issues but have much less hard power when the issue at hand is like (healthcare, crime, immigration). 

 

Here are some cruxes that I don't see addressed as much:

If there is a future with sentient beings living in it, are their lives, on average, likely to be net positive or net negative?

This weighs over all existential risk cause areas. If preventing existential risk is possible, but the future is net-negative, then this intervention may be harmful, due to increasing s-risks.

How much will the governance of developed countries influence the governance of underdeveloped countries?

A country that does not value net welfare at all but instead dominance and power can transfer its values to the rest of the world through popular media and economic dependency. If this happens, it could lead to a future where coordination on issues becomes difficult, if not impossible.

This seems a bit related to the “Pivotal questions”: an Unjournal trial initiative   -- we've engaged with a small group of organizations and elicited some of these -- see here.

To highlight some that seem potentially relevant to your ask:

What are the effects of increasing the availability of animal-free foods on animal product consumption? Are alternatives to animal products actually used to replace animal products, and especially those that involve the most suffering? Which plant-based offerings are being used as substitutes versus complements for animal products and why?

Wellbeing measures/how to convert between DALY and WELLBY welfare measurements on assessing charities and interventions.

Is WELLBY the most appropriate (useful, reliable...) measure [for interventions that may have impacts on mental health]

What is cell-cultured meat likely to cost, by year, as a function of the level of investments made?

How often do countries honor their (international) agreements in the event of large catastrophes (and what determines this?)

How probable is it that cell-cultured meat will gain widespread consumer acceptance, and to what timescale? To what extent will consumers replace conventional meat with cell-cultured meat?

How important is democracy for resilience against global catastrophic risk?

How generalizable is evidence on the effectiveness of corporate animal welfare outreach [in the North] to the Global South?

How much will the US government use subjective forecasting approaches (in the way the DoD does) in the next ~50 years?

"What does 'negative lives' really look like?"

Given its gravity, I find this question underexplored and current tools (QALYs, WELLBYs, YLSSs etc.) either underdeveloped or inadequate. At the same time I think answers should fit into existing frameworks when possible, since these tools often form a cornerstone, or are heavily implied, when doing cause prioritisation.

There are at least two things I would like to know more about:

1. When does the transition from positive to negative lives happen? What characterizes this shift? Some of the above tools have a defined neutral point, but in practice, evaluations around it are somewhat distorted.

2. How far in the negative is it possible to be compared to the positive?

I think it should be possible to at least get some empirical anchors tied to commonly used frameworks, preferably more than one.

(And I have some ideas for how to do so, but it doesn't seem like there are many funding opportunities, so I'd probably want $500 000 for myself to pursue them, or maybe just $50 000 to do some small tests)

I would like to know more about the tractability of economic growth research and interventions in low-income nations. It seems like it has the potential to be much more effective than traditional global health interventions, but there's a lot of uncertainty surrounding it. 

I've done very little research into this, but I would also like to know if economic growth reduces the risk of large-scale anthropogenic violence. Perhaps people living rich and happy lives are much less likely to do things that increase the risk of various global catastrophes occurring. Perhaps the opposite is true. 

Most of these aren't so much well-formed questions, as research/methodological issues I would like to see more focus on:

• Operationalisations of AI safety that don't exacerbate geopolitical tensions with China - or ideally that actively seek ways to collaborate with China on reducing the major risks.
• Ways to materially incentivise good work and disincentivise bad work within nonprofit organisations, especially effectiveness-minded organisations
• Looking for ways to do data-driven analyses on political work especially advocacy; correct me if wrong, but the recommendations in EA space for political advocacy seem to necessarily boil down to a lot of gut-instincting on whether someone having successfully executed Project A makes their work on Project B have high expectation
• Research into the difficulty of becoming a successful civilisation after recovery from civilisational collapse (I wrote more about this here)
• How much scope is there for more work or more funding in the nuclear safety space, and what is its current state? Last I heard, it had lost a bunch of funding, such that highly skilled/experienced diplomats in the space were having to find unrelated jobs. Is that still true?