> More than just being annoying, unsolicited outreach suggests you don’t respect boundaries, which is disqualifying for someone deciding whether to trust you with their money.
This also has negative effects for beyond just the person you have approached abrasively. Anthropic employees talk to each other about inbound they receive! So if you're particularly annoying, you'll not just disqualify yourself from that person but also become known to others as annoying and worth ignoring.
> I’m locked out, so why shouldn’t I just email them anyway?
I think in practice people are just not that locked out of reaching Anthropic staff.
Being legibly excellent can and should apply to just being public about that information, i.e. post it on your website, make a substack, post to the EA Forum, LessWrong etc.
You should also be reachable, i.e. give people a way to contact you that isn't really hard to hunt down from the good public information you're putting out, and isn't annoying/complex to use.
People probably underrate the extent to which being publicly legible + accessible can work to get you considered and generate leads. Of course, this doesn't mean you get funded - you might still be a mismatch for what people are looking for/interested in. But you do stand a much better shot.
(Bonus: it also means that if you do ever get an in, you have something you're immediately able to share with the person, rather than letting that lead go cold!)
---
Finally, another thing that should be obvious but (appears) isn't: Anthropic employees are generally busy folk with little spare time, possibly the worst pitch to them is "can we set up a call to discuss with very little by way of concrete details". The second worst pitch is "here is some Claude slop / half baked poorly thought through plan, give me money". The third worst pitch is "consider engaging with this extremely long resource e.g. read my book, then give me money".
Also, while AI is kinda bad at writing, probably all communications you intend to go to high-value donors should be thought through carefully by hand and written by hand. You should then get Claude to roast your writing, and check for obvious flaws/problems. But AI slop will turn people off very quickly.
I’ve got some advice. If you want to give your tainted money you should give it anonymously— otherwise people should assume you intend to influence what the charity does in favor of your AI company.
I also advise any charity that cares about control over their mission and their integrity to adopt a policy similar to PauseAI US of not accepting money from the AI industry.
Yes, that entire space is pulling punches on Anthropic in hopes of getting their money. The mere possibility that an Anthropic millionaire would donate to AMF— something you are suggesting in this post— is laundering the image of Anthropic and justifying what they do. It makes Anthropic the de facto leader of Effective Altruism.
In case you’re wondering, yes, you are part of this laundering. If your clients really want to AMF to have the money without it being an indulgence/pass for Anthropic, they’ll do it anonymously.
I assume because they feel like a) taking AI donations might lead to those donors trying exert control over their activities; and b) because it is an important signal of their independence from that sort of influence. That is, I guess most people working for frontier AI labs wouldn't donate there in reality, because they don't believe in pausing AI development (although I don't know this), and so the actual cost of not taking donations might not be that high, but it's an important signal to supporters of independence.
But I assume it isn't a general objection to non-anonymous donations because they might benefit the donor's reputation. That is, I assume they would take donations from wealth from other industries that have a bad reputation, like coal extraction or gambling or tobacco, or hereditary wealth that benefited from slavery. (If you have a public policy on this, I'd be keen to educate myself on it. Also worth noting that I said "they" but I should probably write "you", as you are the US founder - I'm just trying not to make this personal.)
That case doesn't really apply to AMF though, where AI donors are very unlikely to want to direct what the organisation does. I think the equivalent there would be AMF taking donations from people who manufacture insecticide or bednets (or, facetiously, who manufacture mosquitos).
So I think the argument that AMF should reject AI donors is wrong, but the argument that PauseAI should reject them is noble.
Okay well I set that policy and, yes, it’s because we are opposing the AI industry and the incentives would be wack. But you asked if it was bad to give money Anthropic windfall to AMF amidst all this hype about Anthropic philanthropy money and, yes, there is a problem with the whole thing. I believe Anthropic has encouraged this hype about all the coming money to keep the space of people who would be most able to criticize them and have the greatest moral duty to stop them hoping to get kickbacks.
Anthropic has already turned EA into its army and it’s using them to take a shot at destroying the world. This money could do some great charity but that’s potentially the bigger effect, and it’s deadly serious.
I do not object to non-anonymous donations in almost any other context. I post publicly about the donations I give even though some people find it crass bc I think it’s something to be proud of. But in this case it’s how Anthropic is using the promise of this money for soft power that makes me say that money needs to be given absolutely without condition or creating perverse incentives (for the AI Safety space, especially) to follow Anthropic.
Fortunately, there is a way to do that, and it’s by giving anonymously.
In PauseAI US’s case people say we should be willing to take the money for the greater good and just not be affected by bad incentives, but generally what they actually mean is that we should like Anthropic and see what they are doing as compatible with our mission, because that’s what it would take to make sense to have a patron relationship with them even indirectly. If the would-be donor won’t give anonymously then that’s a sign that what they wanted was that influence.
> More than just being annoying, unsolicited outreach suggests you don’t respect boundaries, which is disqualifying for someone deciding whether to trust you with their money.
This also has negative effects for beyond just the person you have approached abrasively. Anthropic employees talk to each other about inbound they receive! So if you're particularly annoying, you'll not just disqualify yourself from that person but also become known to others as annoying and worth ignoring.
> I’m locked out, so why shouldn’t I just email them anyway?
I think in practice people are just not that locked out of reaching Anthropic staff.
Being legibly excellent can and should apply to just being public about that information, i.e. post it on your website, make a substack, post to the EA Forum, LessWrong etc.
You should also be reachable, i.e. give people a way to contact you that isn't really hard to hunt down from the good public information you're putting out, and isn't annoying/complex to use.
People probably underrate the extent to which being publicly legible + accessible can work to get you considered and generate leads. Of course, this doesn't mean you get funded - you might still be a mismatch for what people are looking for/interested in. But you do stand a much better shot.
(Bonus: it also means that if you do ever get an in, you have something you're immediately able to share with the person, rather than letting that lead go cold!)
---
Finally, another thing that should be obvious but (appears) isn't: Anthropic employees are generally busy folk with little spare time, possibly the worst pitch to them is "can we set up a call to discuss with very little by way of concrete details". The second worst pitch is "here is some Claude slop / half baked poorly thought through plan, give me money". The third worst pitch is "consider engaging with this extremely long resource e.g. read my book, then give me money".
Also, while AI is kinda bad at writing, probably all communications you intend to go to high-value donors should be thought through carefully by hand and written by hand. You should then get Claude to roast your writing, and check for obvious flaws/problems. But AI slop will turn people off very quickly.
Big +1 to all of this
I’ve got some advice. If you want to give your tainted money you should give it anonymously— otherwise people should assume you intend to influence what the charity does in favor of your AI company.
I also advise any charity that cares about control over their mission and their integrity to adopt a policy similar to PauseAI US of not accepting money from the AI industry.
Interesting take. You think a donation to provide anti-malaria bednets is trying to favour an AI company?
Yes, that entire space is pulling punches on Anthropic in hopes of getting their money. The mere possibility that an Anthropic millionaire would donate to AMF— something you are suggesting in this post— is laundering the image of Anthropic and justifying what they do. It makes Anthropic the de facto leader of Effective Altruism.
In case you’re wondering, yes, you are part of this laundering. If your clients really want to AMF to have the money without it being an indulgence/pass for Anthropic, they’ll do it anonymously.
I'm about as sure as I can be that that isn't why PauseAI isn't taking donations from AI wealth.
Why do you think it is?
I assume because they feel like a) taking AI donations might lead to those donors trying exert control over their activities; and b) because it is an important signal of their independence from that sort of influence. That is, I guess most people working for frontier AI labs wouldn't donate there in reality, because they don't believe in pausing AI development (although I don't know this), and so the actual cost of not taking donations might not be that high, but it's an important signal to supporters of independence.
But I assume it isn't a general objection to non-anonymous donations because they might benefit the donor's reputation. That is, I assume they would take donations from wealth from other industries that have a bad reputation, like coal extraction or gambling or tobacco, or hereditary wealth that benefited from slavery. (If you have a public policy on this, I'd be keen to educate myself on it. Also worth noting that I said "they" but I should probably write "you", as you are the US founder - I'm just trying not to make this personal.)
That case doesn't really apply to AMF though, where AI donors are very unlikely to want to direct what the organisation does. I think the equivalent there would be AMF taking donations from people who manufacture insecticide or bednets (or, facetiously, who manufacture mosquitos).
So I think the argument that AMF should reject AI donors is wrong, but the argument that PauseAI should reject them is noble.
Okay well I set that policy and, yes, it’s because we are opposing the AI industry and the incentives would be wack. But you asked if it was bad to give money Anthropic windfall to AMF amidst all this hype about Anthropic philanthropy money and, yes, there is a problem with the whole thing. I believe Anthropic has encouraged this hype about all the coming money to keep the space of people who would be most able to criticize them and have the greatest moral duty to stop them hoping to get kickbacks.
Anthropic has already turned EA into its army and it’s using them to take a shot at destroying the world. This money could do some great charity but that’s potentially the bigger effect, and it’s deadly serious.
I do not object to non-anonymous donations in almost any other context. I post publicly about the donations I give even though some people find it crass bc I think it’s something to be proud of. But in this case it’s how Anthropic is using the promise of this money for soft power that makes me say that money needs to be given absolutely without condition or creating perverse incentives (for the AI Safety space, especially) to follow Anthropic.
Fortunately, there is a way to do that, and it’s by giving anonymously.
In PauseAI US’s case people say we should be willing to take the money for the greater good and just not be affected by bad incentives, but generally what they actually mean is that we should like Anthropic and see what they are doing as compatible with our mission, because that’s what it would take to make sense to have a patron relationship with them even indirectly. If the would-be donor won’t give anonymously then that’s a sign that what they wanted was that influence.