CCF: Billions are coming. Still no plan to give well?

TLDR: Everyone’s talking about what the money could do, but few about how to decide where it goes.

This post is part of the new series of articles on cross-cause giving and the new wave of philanthropy. Stay tuned to the EA Forum and our Substack for the latest takes on topics such as giving now vs. later, common pitfalls in cause prioritization, and other crucial considerations from the Cross-Cause Fund (CCF) team at Rethink Priorities.
  
Amid all the discussion of the coming wave of AI-driven philanthropy, one question gets surprisingly little attention: how should this money be allocated across causes, and on what basis?

The stakes are large. One back-of-the-envelope calculation estimated that yearly philanthropic spending could increase by $37 billion to $100 billion, a 6–17% jump. About two-fifths of that ($15–40 billion) will likely come from Anthropic founders and employees, most of whom are likely to give in a way influenced by effective altruism, trying to maximize impact per dollar. (The remaining three-fifths might come from the OpenAI Foundation.) That is fifteen to forty times the $1 billion moved by effective giving organizations in 2024, every year.

A number of people have raised concerns about this surge in giving. Sjir Hoeijmakers is worried that the money will be spent on charities that make donors feel good, rather than on charities that most effectively improve the world. Nan Ransohoff is worried about scaling up nonprofit infrastructure to handle this much giving. Lewis Bollard is worried about negative effects on the movement culture, e.g., collaboration between nonprofits, maintaining shared values as the movement expands, and the discipline to spend each dollar in the best way to improve conditions for animals. Morgan Fairless worries that charities spend a lot of time preparing for a wave of philanthropy that never materializes.

All of these concerns are reasonable, but they sit downstream of a more fundamental one. Too often, the allocation question is answered by default: donors lean on a trusted advisor’s instincts, a cause they feel connected to, or the assumption that there will be so much money that everything worth funding will get funded anyway. Each of these is a decision about tradeoffs, made without being examined, e.g., between humans and animals, present and future generations, certain benefits and uncertain ones, giving now and giving later. A flood of money doesn’t make those tradeoffs go away. It raises the stakes of getting them wrong.

Will there be enough opportunities to fund? Spoiler: Yes.

I have seen some people argue that AI-related donations will be so large that they will saturate all effective cause areas, so we don’t need to worry about doing comparisons between different uses of the money.

First of all, this is a very flip attitude to take about human and animal lives. It amounts to saying, “it doesn’t matter if I send tens of millions of dollars to the wrong place! Eventually, someone will fund the right things. Not me, but you know, someone.”

Second, this amounts to saying that the AI-related wave of philanthropy will be sufficient to solve all world problems. I will believe this happens when it happens. The American philanthropic sector spends $600 billion every year, and yet as far as I can tell, the world still has problems. It is possible to catastrophically mismanage enormous sums of money, such that tens of millions of dollars are spent on giving a concert hall somewhat better acoustics, while opportunities to save a life for ~$3000 go unfunded. I’d argue this type of mismanagement is the norm, and is likely to continue, even among people trying to do better, without careful thought about how to allocate across different areas.

While AI-related donations will allow us to do a lot of good, their impact is likely limited. We should not prepare for extreme ambitions and unlimited resources, or as Neil Dullaghan put it, “We’re not talking about plans to absorb trillions of dollars to buy Tyson Foods and build cultivated meat on the Moon.”

Determining if there’s something left worth funding depends on thinking about tradeoffs

Whether you know it or not, when making donation decisions you are always making tradeoffs. There’s no such thing as a “saturation point” at which point important causes would no longer be able to spend money effectively, or if there is, it’s so high that we have no realistic prospect of reaching it any time soon. Even if you look at just one cause area, like Global Health, below the GiveWell top charities are charities that just barely missed their bar. Below those charities are others that are somewhat worse or more thinly evidenced. At some point, if you can’t think of anything else to do with the money within that cause, you could work with GiveDirectly to create a universal basic income for everyone in the entire world.

But you can also look more broadly. If you care about the suffering of sentient beings, the ladder doesn’t end with humans. Corporate campaigns to reduce the suffering of farmed chickens have been estimated to spare hundreds of hens from cage confinement per dollar, which is plausibly orders of magnitude more cost-effective, in suffering averted, than even the best global health interventions, if you’re willing to give chickens some moral consideration. Beyond chickens, there are billions of farmed fish and shrimp whose welfare receives almost no philanthropic attention. These are areas where a marginal dollar can still buy improvements so cheaply that they, too, clear cost-effectiveness bars some global health interventions can’t, assuming you give these species some moral concern. And beyond the present entirely, there are efforts to reduce catastrophic risks from engineered pandemics, nuclear war, or advanced AI, where if you believe the future might be big, the numbers at stake are so large that even small changes in probability matter enormously.

Long before any of these areas “fills up,” you’ll hit the limits of your budget. The question is never whether there’s something worth funding; it’s which of the many worthwhile things you’ll fund first.

It’s not obvious that something a few steps down from GiveWell’s funding bar deserves no money, and it’s not obvious that they deserve to be flush with cash. But if you don’t think about what you’re doing, you’re probably going to do one or the other more-or-less at random.

“Effectiveness” isn’t an on/off switch

Another issue with thinking there’s some clear saturation point after which there’s nothing to fund is that it seems to stem from a belief within the effective altruism (EA) community that, once you run out of options the community currently considers high-value, you can do whatever you want with your money. Inside or outside of EA, people often implicitly divide charities into approved (”effective”) and unapproved (”ineffective”). They see all charities that don’t currently have the stamp of approval as a homogeneous mass of Ineffective Charities. If all the Effective Charities have their funding gap filled, you can do whatever you want with the rest.

But, in reality, these charities range widely in effectiveness. Some are net-harmful, while others are as good as or better than charities currently considered “effective” by the EA movement. If you ought to spend your donations effectively, which I do think you should, then your obligation doesn’t go away when you have run out of already vetted and investigated charities. You can and should vet and investigate new charities! This assumption doesn’t make much sense, which I think is why people rarely make the argument explicitly.

Making tradeoffs between apples and oranges

Philanthropy presents a number of thorny ethical issues which we don’t know how to resolve. We don’t know how to trade off between humans and chickens, or between present people and future people, or between improved health and increased income. We don’t know how risk-averse we ought to be. We don’t know how much certainty to put into moral theories that, for example, don’t allow you to aggregate many small health benefits to be the equivalent of saving a life. We don’t know whether you should combine moral theories or go with your best guess, and if you do combine them, we don’t know how.

I’ve been talking — and many people talk about this issue — like you can line up every giving opportunity on a single scale and proceed down the list. But this is a mistake. Doing good isn’t like stock valuation or investment in a for-profit company. When you invest in a for-profit company, you know what your goal is, i.e., to earn as much money as possible. But when you do good, you don’t. Sincere people trying to improve the world as much as possible come to wildly divergent views about what the best thing to do is.

The effective altruism movement is pretty good, though not perfect, at addressing empirical uncertainty through monitoring and evaluation programs, RCTs, and charity evaluators such as GiveWell and the EA Funds. But the decisions that most change where money goes are usually not only empirical; they’re also moral. It is rarely the case that one person donates to cage-free egg campaigns and another donates to malaria net distribution because they have different reads of the empirical evidence. Instead, the person who donates to cage-free egg campaigns likely values chickens more than the person who donates to malaria net distribution, or they value the certainty of the benefit to humans less than the increased potential size of the benefits to chickens.

In order to treat doing good like a stock valuation, you have to trust that you can weigh evidence in ethics and come to a robust decision. But we aren’t that good at ethics! The evidence, in many cases, consists largely of thought experiments. You shouldn’t have the same confidence in your answers to thought experiments that you have in empirical, real-world results that you can point to.

Many people seem to assume that a flood of money solves these problems, because you can just fund everything. But you can’t realistically fund everything. A flood of money doesn’t make the problem easier; it just raises the stakes of solving it. No one has a real model of how best to spend billions of dollars, because no one has a real model of how best to spend millions or thousands of dollars.

Think you already know how to distribute effectively? What’s your actual plan?

The best way to see how unprepared we are here is to ask people what their practical plan is: what do they think the distribution of money should be and why they think it’s the best. Often, they will look blank or stutter through an explanation, making it clear they don’t yet know.

Much of the poor reasoning around the third wave of philanthropy may stem from being overwhelmed by the difficulties of making these decisions. It’s easy to give up and say that all the reasonable choices are equally good so you can give to whichever ones you feel like. However, these problems are hard, but not impossible to tackle. You can make real progress on giving in a principled way.

In my view, to give effectively, you need at least:

• Up-to-date information about how charities you might donate to plan to spend money.
• Up-to-date information about where other donors are spending money.
• Accurate models of how many benefits of which kinds come from that spending.
• A clear way of thinking about tradeoffs between different kinds of goods, including philosophically confusing concepts like moral weights, risk aversion, and aggregation of small goods.

Sometimes, we can be tempted to think that we don’t need a complex, explicit model; we’ll just use our best judgment. But any time we make a decision about how to allocate our donations, we’re, in fact, using a model, at least tacitly. Making the model explicit allows us to question and reconsider the assumptions we make, figure out which considerations actually matter for the final decision, and to learn how sensitive our conclusions are to aspects of the model we’re uncertain of. Often, if you don’t make your model explicit, your best judgment simply reflects your biases (towards humans, towards charities you have an emotional connection to, towards problems that feel vivid...).

At Rethink Priorities, we built the Cross-Cause Model, and the accompanying Cross-Cause Fund, to take these questions on directly. It brings together empirical estimates of what interventions achieve with explicit, adjustable assumptions about the hardest moral questions: how to weigh different beings, how much risk to accept, and how to act when moral theories disagree. It draws on years of research across global health, animal welfare, and catastrophic risk. It isn’t perfect, and it isn’t the only possible approach. But we believe it is currently the most comprehensive all-things-considered philanthropic tool available for allocating across causes.

Whatever approach you use, though, you need one. The more money at stake, the more it matters that your decisions rest on reasoning you can examine, question, and improve.

In the coming weeks, we plan to address various concerns about philanthropic giving, like:

• Giving now vs. later
• Giving in the face of new AI trajectories
• Cause prioritization done wrong and how to prevent it
• How to avoid giving disasters
• And many more.

Stay tuned.
 

Acknowledgements

Thank you, Ozy Brennan, for help with writing this post, and Ula Zarosa for editing.

Parts of this content were prepared with the assistance of AI tools (such as Claude and Gemini), which we use to improve efficiency and readability. All outputs are supervised, reviewed, and fact-checked by RP staff, who remain responsible for the final content.