Since somebody nudged me to reply to this. I didn't find this reply very convincing.

First, I agree that work pulling people into AI safety is/will be more time-consuming. Other than that, though,

I think that the ratio of grantmakers:people in AIS isn’t that informative for answering the question, “is the number of grantmakers a bottleneck”

This seems off to me. Surely you'd agree that if the ratio were 1:1 or something, we would say "ok, some people who are currently grantmaking need to be doing direct work". Not everyone can just be funding things.

The better ratio is presumably something like, “people who could be working in AI safety/governance if they got funding: grantmakers”

I mean, maybe that's another one to consider, but it still feels less relevant. A lot of people think that the denominator in this case is infinite/(nearly) all humans.

We currently feel more constrained by evaluating or creating new funding opportunities than directing $ to them

I actually don't think this necessarily implies that we should get a lot more people into grantmaking. This could just mean that there is a lot of money available and that grantmakers should move into roles on the ground (direct work)

Hi Michael, I liked this comment a lot, strongly upvoted.

If I were to give a few suggestions off the top of my head:

1. It often feels like AIS's answer to get what they want, especially faster, is to simply throw more money at the problem. Salary raises, perks, nicer spaces, fancier events, etc. I think money is perhaps one knob to turn, but I'd like to see other knobs turned as well: creativity, working a lot longer/harder (sacrifices of free time), etc. Importantly, I don't want these to be inane (spend 1 week per year in Malawi living like the poorest people on Earth). If people had the sense that AIS people got paid well but were really burning the midnight candle, I think the common man (or parts of EA which aren't as flush with cash) would have less disdain for them.
2. Abraham kind of mentions this, but maybe any kind of costly signal (or rather, hard-to-fake signal).
3. Don't spend money just because you can. It often feels like money in AIS is spent sort of recklessly, because it's in excess rather than "yeah, there's an actual need for this thing",  and thus it doesn't really matter.
4. Don't make "money moved" the metric for AI safety. It often seems to be. And it's probably convenient. But it's pretty easy to move money. It's harder to move money well.
5. Just a lot more public writing of thought processes. It seems that more and more AIS is done in private Google Docs and Slack channels, and just occasionally, clearly public-facing posts are made, and it's very clear that "this is what the masses get to know, the professionals will manage the real stuff". The original name of Coefficient Giving was "Open Philanthropy". Supposedly, it was changed to not be conflated with OpenAI, but a more cynical view (which I don't believe) would be that it was done to be more accurate in the shift that had taken place and will continue to take place. There's a similar case with Edward Snowden where he made the point that by default, everything in the government is classified even though it ought not be. It often feels like CG sees itself as the "steward of AIS," where everyone else should do the thing they are told.

Perhaps another thing I'd like to see is just frankly less pompousness around "I am one of the most important/effective people in the world, doing great things for the world, having lots of impact". I get the sense that many think it's true, but it's a bit hard to prove and comes across a little thick when it's paired with what looks on the outside to be a rather comfortable-looking job with great perks/events.

This piece by Lincoln Quirk really comes to mind

Hi Hans, I appreciate this a lot. I just feel i should say, I am by no means poor. I dont spend much money but I have some investments (private and public) that have done really well and job offers that make me confident ill be fine no matter what. My worst case scenario is I do some boring market making or investment banking. 

I've seen a lot of posts that we need a lot more AI safety grantmakers. I feel like I want to do a bit of rough math and just see if that's the case. There is this estimate for the number of FTEs in AI safety by Stephen Mcaleese from Sept 2025 and 2022. Let's extrapolate exponential growth and say there are ~1400 FTEs on AI safety right now. Let's also assume from Julian Hazell's post that there are ~50 full-time AI safety grantmakers (though I think it's probably a bit more than that, given CG, Astralis, Astera, Longview, SFF, independent grantmakers, FLI, UK AISI, ARIA, AISTOF, Navigation Fundpeople at Schmidt, Macroscopic, etc., LTFF, Bluedot grants, Manifund, Tarbell, etc.).

From what I know about CG and other grantmakers, the people there are quite talented, and I would speculate are more talented than the average grantee.

Right off the bat, that means that right now, about 28 FTEs are working in AIS per grantmaker. Not to mention, a lot of the people who work full-time in AIS are working at frontier labs or other for-profit companies like Goodfire or in government (like UK AISI, CAISI), who don't need grantmakers to evaluate/fund their work. But we can ignore all those and just stick with the 28 FTE number.

I think I would expect the average grantmaker to be able to handle more than that, especially since an average organization usually has ~10 FTEs on average (I just asked Claude), and I expect a typical grantmaker to handle much more than 3 grants.

Also, I suspect a lot of grants look a lot more like renewals, and so don't need nearly as much review. For example, I'd expect grants to MATS and Redwood to look a lot more like reviewing their plans and signing off on them.

What am I missing?

I just wanted to say, Phil's discussion with me in DMs has been very good about this and im going to be testing this too with some people.

I think he wrote this post off the cuff but this has been tremendously underdiscussed

This just came to mind: the reason that it's the wrong way to go about solving problems is that you want to solve the largest problems (well, per resource) and not just solve any random problem. Like, there is a problem that my shoes are currently untied, and I don't want to bend down or spend 10 seconds to tie them, but it's not very important.

So if you want to solve the most important problems, you should start with the problem and then work backwards for what solutions you might wish existed. I think the mere fact that people often talk about forecasting as the solution they are seeking to apply, whether that be Sentinel or whoever, is evidence that things are going wrong.

This just came to mind: the reason that it's the wrong way to go about solving problems is that you want to solve the largest problems (well, per resource) and not just solve any random problem. Like, there is a problem that my shoes are currently untied, and I don't want to bend down or spend 10 seconds to tie them, but it's not very important.

So if you want to solve the most important problems, you should start with the problem and then work backwards for what solutions you might wish existed. I think the mere fact that people often talk about forecasting as the solution they are seeking to apply, whether that be Sentinel or whoever, is evidence that things are going wrong.

One thing I'd flag is that models are extremely good at telling who is prompting them, and this leads to them being sycophantic, in very subtle ways. I'm not quite sure how they do it, but I've seen this in multiple instances.

For starters, you'd want to lobby for external safety testing on internal models.  You'd want to make sure external safety researchers had access to the models. You'd want certain reporting, etc.

I think he would include a lot of people who work at Anthropic, for example, on pre-training, some of whom went through MATS or something.