Associate Program Officer on Coefficient Givingâs Short Timelines Special Projects team.
đ¸ GWWC Pledger
Michael Townsendđ¸
Karma: 2,652
If itâs average future that still could justify a 1x bar, depending on what weâre averaging over.
I donât think it does. Itâs conceptually coherent for an organization to have a very high average cost-effectiveness while also having a marginal cost-effectiveness below 1x. For this reason, I donât think you should have a âbarâ for average cost-effectiveness. (You might be making the point that if the average cost-effectiveness is above 1, then you are better off making the grant than burning the money, and so it clears a bar in that sense, but itâs not clear itâs worth making the grant vs making a potentially much smaller grant, and so itâs not a helpful âbarâ in the sense the term is usually used.)I agree with the concerns about uncertainty, displacing less-effective charities, and counterfactuality. But Iâd rather see attempts to adjust the estimate for that rather than ~âweâre saying 6x but not really, probably lower after considering thisâ. This will help avoid temptations towards soldier/âpromotion mentality, and make it more comparable to other estimates.
Sure, but these are hard to account for. I agree itâs better to adjust the model when itâs possible, but youâll still be left with a model that has a tonne of uncertainty.
(RE âopportunity cost of the labor of the employees who could otherwise do impactful work or eartn to give, etcââif EA people are putting in free labor into these efforts, that should also be factored into the cost estimates, naturally, not just the direct CG investment.)
Yep! I wasnât trying to suggest you shouldnât account for that.
Saying âdeliverâ to me present tense implies âdeliver and will continue to deliverâ, suggesting the marginal returns should be comparable.
This could still be (and Iâd guess is?) referring to the past and expected future average cost-effectiveness.
I also think that itâd be pretty reasonable to have a bar higher than 1x. (I donât know what CGâs bar actually is.) There are many contentious choices you make when coming up with a multiplier â e.g., how do you discount future donations, how do you discount donations to less cost-effective charities, do you adjust for the opportunity cost of the labor of the employees who could otherwise do impactful work or eartn to give, etc. Thereâs also just a huge amount of uncertainty in various places, especially around counterfactuality. So given all that, I think itâz reasonable to just zoom out and think: hmm, this intervention looks great overall, but I think the multiplier model isnât robust enough to justify further support of organizations that it estimates only have a 1.1x multiplier.
(Flagging I work at CG, but not on this team!)
Curious what you think of the Giving What We Can impact evaluation? I helped write the first one, which claimed that over the 2020â2022 period it had an average multiplier of 30x. I think the more recent one claimed it had a 6x multiplier.
If an investment in an organization can yield anything over 1x in counterfactual donations, this would be worth funding.
I think thereâs an important distinction between average and marginal effectiveness â the above claim seems true in the abstract for orgs that can move >$1 on the margin but not on average. And IMO itâs much harder to estimate marginal cost-effectiveness, because itâs forward looking, whereas historical average cost-effectiveness is a bit easier.
Thanks for the kind words :) and appreciate the concreteness of these suggestions.
I think many of these seem to have a common thread of: treating money as a valuable resource, viewing profligacy as costly (both financially and to the soul), seeing signals of dedication and altruism as especially valuable, and generally being pro transparency.
I agree with all of these, though I think itâs difficult to know how to weigh them against the things they sometimes trade-off against. For instance, if youâre deciding whether to fund a large and potentially very impactful grant opportunity that involves high salaries, where the details of the grant are sensitive, it feels unclear to me how much impact there has to be on the table to justify the high salaries and discretion the grant involves. Iâm pulled towards posts like this, though I likewise feel some pull towards the ideas Will discussed in the EA and the current funding situation (or at least my memory of it, which is that in some cases itâs worth setting aside our preference for an ascetic aesthetic when resources are more plentiful and the worldâs problems are urgent.)
The thing Iâm especially curious about is exactly how power and influence can start corrupting oneâs thinking, and what ways there are of avoiding that (to the extent there are any).
Thanks for writing this.
I started my EA journy at Giving What We Can, encouraging others to pledge to give 10% of their income to effective charities. I got into allt his because I thought it was basically insane that you could save a life for $5k. I still do, but now Iâm working in AI, where thereâs significant amounts of funding available, powerful people involved and interested, the stakes are existential, I feel like my scope sensitivty cannot keep up anymore. So lot of what youâre saying resonates.
I think EAâs failure to grapple with the corrupting influence of power is among its greatest failures.
I both agree with this in the abstract and also feel pretty clueless about what grappling with this would look like. I worry that one way to grapple with it is to try to avoid it and flee to a place with less moral ambiguity, but also at the cost impact. Iâd be interested in your thoughts on what it would look like for people who are wielding that power to take seriously its corrupting influence.
Fair enough â I think I was trying to say something along the lines of âgoing through any specific example invites a lot of genuinely thorny and difficult questions about counterfactuality/âsign of impact/âattribution to EAâ (and again many of these are hard to discuss on a public forum) but I think zooming out, you can see EAs fingerprints in various important places. I think this leads to an overall common-sense perspective that EA has helped improve the situation.
Also, I agree I pointed to work in the middle of the ToC chain, but that seems kind of reasonable to me given that AI is currently not that powerful and not really that scary. AI hasnât yet been capable of causing a disaster, so itâs not really possible to have prevented one (yet).
On the specific example of Redwood Research is doing a lot of really valuable safety work. I think pioneering Control has been a fairly useful accomplishment, and I suspect if someone wanted to dig into the details, theyâd find that it was fairly counterfactual.
I guess as you disclaimed might be the case up front, I donât think these are the strongest or most informed examples of EAs impact on AI safety.
In many of cases of such impact, one can quibble about many things:Whether that impact was clearly positive, or whether it had some kind of indirectly negative harmful effect, most commonly via speeding up AI development. See Paul Christianoâs reflections on the impact of Reinforcement Learning with Human Feedback as an example.
The counterfactuality and persistence of the impact â e.g., like you said for many of these, would this have happened (eventually) anyway?
How attributable that was to EA (and unfortunately in some cases, due to EA having a toxic brand in many places, itâs actually best if it is not that attributable to EA).
And last âDoes any of that matter? All of EAs impact â for better or worse â has been its influence on Anthropic.â
Yet, I think taken as a whole, I think EA has punched above its weight in many ways with respect to making AI go well. Itâs led to:
More and better staffed AI safety/âsecurity institutes
A richer non-profit ecosystem of safety research (like Truthful AI, FAR AI, Redwood Research, etc.)
More and better staffed third-party evaluations, auditing, and science (METR, AVERI)
Large amounts of field-building that encourages talented people to work on making AI go well (MATS, BlueDot, 80k)
A significant amount of policy advocacy and public communications about AI risk.
Probably other examples, too.
A lot of the effort to make this happened relied on EA motivated people willing to take lower paid or less glamorous jobs.[1] While some specific organizationsâ or research or policy wins or public communications would have happened otherwise, but some wouldnât, and even still, happening earlier is still better.
I started out in EA caring about global health, and my first EA job was as a Researcher at GWWC. Even after becoming pretty convinced by AI risk and longtermism, I was still fairly sympathetic to concerns like âAI Safety alienating peopleâ. For instance, I was pretty against 80,000 Hours becoming explicitly focused on longtermism, and also pretty skeptical /â worried about its pivot last year into leaning even more into AI. Now, looking at just how fast AI progress is developing, how much there is still be done to make it go well, and how valuable (I think) EA has been to date, I think I got a lot of that wrong.
- ^
And of course, in some cases, they happened to get pretty well-paid jobs that ended up being fairly glamorous (even if they werenât in the beginning). I donât think that undermines the impact much. I donât really begrudge the quant finance folks who give >50% of their income to charities, even if theyâre still pretty rich at the end of the day.
First, OpenAI may be successful in turning a large profit based on continued marginal improvements in AI systems, so that their company continues getting much more valuable, far faster than 5% annual growth, meaning that the endowment would grow in value, that is, on net accumulating rather than distributing wealth. In this case, OpenAIâs equity will appreciate greatly; it would be irresponsible for the nonprofit not to try to spend large parts of that increased value.
In this scenario, wouldnât it be much better if the non-profit didnât spend its money now? By holding onto the money now, itâd have much more to give later. Put another way: imagine if the grantees receiving the money were asked âwould you prefer $100 today or $10,000 in 6 years?â many would take the latter.
One frame that might make this argument more compelling is that if OAI ends up building AGI and ends up having astronomical value, then the foundation is sitting on humanityâs endowment. Spending it down now before itâs realized its value could be very costly.
Thanks for this post!
Curious what you think of the following objection: AI character work has to grapple with the question of whether we want AI systems to be âobedientâ or âethicalâ. Yet, it seems non-obvious which is better for the long-term future because training them to be ethical might make alignment risk worse, and training them to be obedient makes misuse and other risks worse. So if youâre uncertain about which is better â which I think you probably should be â then on expectation, the impact of affecting AI character is reduced proportional to how uncertain you are. I expect there are some ways you can affect AI character work that doesnât have this âobedience vs ethicalâ structure, but I also suspect that the argument made in this report about the importance of such work apply less to these interventions.
Hey Arepo!
Last I heard it was something like 10% of their GCR budget.
I donât think thatâs right â CG gave $400m to GHW in 2025, and to get a sense of what % that might be, Alexander Berger (CEO of CG) shared that overall âCoefficient Giving directed over $1 billion in 2025â in his recent letter.
CEA: now EA-adjacent!
It seems pretty hard to argue with the idea that:
Youâre more informed about what youâre giving up if you take the 10% Pledge after having worked for 1â5 years,
and itâs generally a good thing to be informed before you make life-long commitments!
But even still, I think being an undergrad at university is a pretty great time to take the 10% Pledge. Curious what you/âothers think of the arguments here.
One thought is: why should the default be to keep 100% of your income? The world seems on fire, money seems like it can do a lot of good, and the people weâre talking about pledging are likely pretty rich. I think in almost all cases, if youâre among the richest few percent of the world, itâs the right thing to do to give 10% (or more) of your income to effective charities. And I think that, even while youâre an undergrad, thereâs a good chance you can know with very high confidence youâre going to end up being among the richest few percent â assuming youâre not there already.
An analogy could be raising children as vegetarians/âvegans. I think this is a totally justified thing to do, and I personally wish I was raised vegetarian/âvegan, so that I never craved meat. Some think that itâs not fair to impose this dietary restriction on a child who canât make an informed choice, and itâd be better to only suggest they stop eating meat after they know what it tastes like. But given that eating meat is wrong (and itâs also an imposition on the child to make them eat meat before they know that it is wrong!) I donât think being an omnivore should be the default.
Thereâs admittedly an important disanalogy here: the parent isnât making a lifelong commitment on behalf of the child. But I think it still gets at something. At least in from my own experience, I feel like I benefited a lot from taking the 10% Pledge while at uni. If I hadnât, I think thereâs a good chance my commitment to my values could have drifted, and on top of that, every time I did give, itâd have felt like a big, voluntary/âoptional cost. Whereas now, it feels like a default, and I donât really know what itâs like to do things another way (something I value!).
That said, I think itâs pretty bad if anyone â but perhaps especially undergrads/âpeople who are less sure what theyâre signing up to â feels like they were pressured to make such a big commitment. And in general, itâs a rough situation if someone regrets it. But Iâve previously seen claims that undergrads pledging is bad, and should be discouraged, which seems like a step too far, and is more what I had in mind when I wrote this (sorry if thatâs not really making contact with your own views, Neel!).
Just speaking for myself, Iâd guess those would be the cruxes, though I donât personally see easy-fixes. I also worry that you could also err on being too cautious, by potential adding warning labels that give people an overly negative impression compared to the underlying reality. Iâm curious if there are examples where you think GWWC could strike a better balance.
I think this might be symptomatic of a broader challenge for effective giving for GCR, which is that most of the canonical arguments for focusing on cost-effectiveness involve GHW-specific examples, that donât clearly generalize to the GCR space. But I donât think that indicates you shouldnât give to GCR, or care about cost-effectiveness in the GCR space â from a very plausible worldview (or at least, the worldview I have!) the GCR-focused funding opportunities are the most impactful funding opportunities available. Itâs just that the kind of reasoning underlying those recommendations/âevaluations are quite different.
(I no longer work at GWWC, but wrote the reports on the LTFF/âECF, and was involved in the first round of evaluations more generally.)
In general, I think GWWCâs goal here is to âto support donors in having the highest expected impact given their worldviewâ which can come apart from supporting donors to give to the most well-researched/âvetted funding opportunities. For instance, if you have a longtermist worldview, or perhaps take AI x-risk very seriously, then Iâd guess youâd still want to give to the LTFF/âECF even if you thought the quality of their evaluations was lower than GiveWellâs.
Some of this is discussed in âWhy and how GWWC evaluates evaluatorsâ in the limitations section:
Finally, the quality of our recommendations is highly dependent on the quality of the charity evaluation field in a cause area, and hence inconsistent across cause areas. For example, the state of charity evaluation in animal welfare is less advanced than that in global health and wellbeing, so our evaluations and the resulting recommendations in animal welfare are necessarily lower-confidence than those in global health and wellbeing.
And also in each of the individual reports, e.g. from the ACE MG report:
As such, our bar for relying on an evaluator depends on the existence and quality of other donation options we have evaluated in the same cause area.
In cause areas where we currently rely on one or more evaluators that have passed our bar in a previous evaluation, any new evaluations we do will attempt to compare the quality of the evaluatorâs marginal grantmaking and/âor charity recommendations to those of the evaluator(s) we already rely on in that cause area.
For worldviews and associated cause areas where we donât have existing evaluators we rely on, we expect evaluators to meet the bar of plausibly recommending giving opportunities that are among the best options for their stated worldview, compared to any other opportunity easily accessible to donors.
First just wanted to say that this:
In my first year after taking the pledge, I gave away 20% of my income. However I had been able to save and invest much of my disposable income from my relatively well paid career before taking the pledge and so had built up strong financial security for myself and my family. As a result, I increased my donations over time and since 2019, have given away 75% of my income.
...is really inspiring :).
Iâm interested in knowing more about how Allan decides where to donate. For example:I currently split my donations between the Longview Philanthropy Emerging Challenges Fund and the Long Term Future Fund- I believe in giving to funds and letting experts with much more knowledge than me identify the best donation opportunities.
How did Allan arrive at this decision, and how confident does he feel in it? Also, how connected does Allan feel with other EtGâers who are giving a similar amounts based on a similar worldview?
Just on this point:
Relatedly, it feels like this is not what the username field is for. If Iâm interacting with someone on some topic unrelated to my advocacy it feels intrusive and uncooperative to be bringing it into the conversation
I think this argument might have a lot of power among folks who tend to think of social norms in quite explicit/âanalytical terms, and who put a lot of emphasis on being cooperative. But I suspect relatively few people will see this as uncooperative/âintrusive, because the pin and the idea itâs advocating are pretty non-offensive.
Luke, thank you for everything youâve done for GWWC and the world.
I donât think many people get to meet someone with such extraordinary levels of care, both for those far away in space/âtime, and loved ones nearby. While I was working at GWWC, Lukeâs most common reason to take a day off was to help someone move house. Luke, your kindness, integrity and commitment are contagious â even with you no longer at the organization, those virtues will stay with GWWC in large part because of how you demonstrated them.
Hereâs a graph of new 10% Pledgers since Luke joined GWWC.[1]
- ^
Courtesy of Claude⌠though my critical feedback for it is that it makes it look like 2024 hasnât happened yet, and that Luke joined in 2019. Both false!
- ^
Unfortunately, I canât see that option â it just displays my email.
Thanks for sharing this, and more importantly, for writing it. From my perspective, this is the best reporting on AI that Iâve seen. Iâve shared it with previously ultra-sceptical friends, and had an uncharacteristically positive response.
The example you gave is about marginal cost-effectiveness (we spend â$1 on adsâ). I agree that then, in this abstract/âidealized case, you should spend the $1 on ads. I think all the uncertainty you would realistically have makes it less obvious, though.
But average cost-effectiveness would be more like, we spent $1,000,000 on an organization that did a bunch of different activities, and we think that led to $1,500,000 counterfactually going to charity. This seems good on average, but thereâs a further question of whether we should give another $1 to the organization. And I think that the 6x figure of the orignal post is referring to average cost-effectiveness (âour current grantees deliver an average adjusted return on donations of 6x across our effective giving portfolioâ). This is at least conceptually coherent with the bar for the marginal $ being closer to 1x.
I think you might find the GWWC impact evals interesting, they go into an enormous amount of depth on all these issues.