I’m super encouraged that you think a large amount of money will flow in to good causes from Anthropic employees. I really hope you are right, although I think there’s likely to be moderate-heavy disappointment on the Anthropic/Giving front.
Maintaining radical generosity and the altruistic values you used to have on the path to riches is really difficult. Most people unfortunately will become far more selfish, and rationalise keeping money they previously intended to give away. Rare cases like Moskowitz have shown its possible, but there are surprisingly few cases of young people (under 40) giving away millions after a cash windfall, let alone tens of millions. I’m not blaming anyone here at all, just stating an observation.
On the other hand I’m encouraged by cases like @richard_ngo that do distribute good amounts of riches that come in. So its difficult but possible!
This seems quite correct! There quite a few open questions in my mind.
1) What is the chance Anthropic EA’s either aren’t interested in donating or will spread their donations out across significant time? If they are EA, it seems unlikely that they will struggle to understand the “donation timing” case.
2) What percent of the IPO do we expect to be donated? To which cause areas?
3) What is our estimate for how logarithmic the utility functions of common EA charities are?
4) Isn’t AI just predicting the next word? Why would Anthropic be able to make any money from this?
What is our estimate for how logarithmic the utility functions of common EA charities are?
The argument is less about how the value of the marginal dollar falls off, but instead about how smoother funding is usually much more valuable to projects. Imagine I’m trying to decide between donating $100k today and $100k in a year, or $200k today. If I expect everyone else together will be donating $1M today and $10M in a year, then I should probably pick the $200k today. The idea is that the charity can probably much more productively use $1.2M today and $10M in a year than $1.1M today and $10.1M in a year.
These are good questions. I don’t know whether (4) was intended as a joke, but I think it’s sincerely a good question. There hasn’t been any compelling use case so far of LLMs making money for the end customers, although that doesn’t stop Anthropic from getting a huge speculative valuation, and it doesn’t necessarily stop employees from selling some of their Anthropic equity in the meantime.
In some senses I really couldn’t blame them for spreading out donations with that kind of windfall. I’d personally just donate the maximum amount per year that I could still write off on taxes which is 60% of annual money income or 30% equivalent income if doing direct stock transfers.
I’m super encouraged that you think a large amount of money will flow in to good causes from Anthropic employees. I really hope you are right, although I think there’s likely to be moderate-heavy disappointment on the Anthropic/Giving front.
Maintaining radical generosity and the altruistic values you used to have on the path to riches is really difficult. Most people unfortunately will become far more selfish, and rationalise keeping money they previously intended to give away. Rare cases like Moskowitz have shown its possible, but there are surprisingly few cases of young people (under 40) giving away millions after a cash windfall, let alone tens of millions. I’m not blaming anyone here at all, just stating an observation.
On the other hand I’m encouraged by cases like @richard_ngo that do distribute good amounts of riches that come in. So its difficult but possible!
This seems quite correct! There quite a few open questions in my mind.
1) What is the chance Anthropic EA’s either aren’t interested in donating or will spread their donations out across significant time? If they are EA, it seems unlikely that they will struggle to understand the “donation timing” case.
2) What percent of the IPO do we expect to be donated? To which cause areas?
3) What is our estimate for how logarithmic the utility functions of common EA charities are?
4) Isn’t AI just predicting the next word? Why would Anthropic be able to make any money from this?
The argument is less about how the value of the marginal dollar falls off, but instead about how smoother funding is usually much more valuable to projects. Imagine I’m trying to decide between donating $100k today and $100k in a year, or $200k today. If I expect everyone else together will be donating $1M today and $10M in a year, then I should probably pick the $200k today. The idea is that the charity can probably much more productively use $1.2M today and $10M in a year than $1.1M today and $10.1M in a year.
EDIT: added something like this to the post
These are good questions. I don’t know whether (4) was intended as a joke, but I think it’s sincerely a good question. There hasn’t been any compelling use case so far of LLMs making money for the end customers, although that doesn’t stop Anthropic from getting a huge speculative valuation, and it doesn’t necessarily stop employees from selling some of their Anthropic equity in the meantime.
Note that some of the founders have pledged to donate 80% of their equity.
All of the founders, actually!
I think maybe that wasn’t public until 2026-01 with Dario’s “All of Anthropic’s co-founders have pledged to donate 80% of our wealth”?
iirc it was first publicly mentioned in dec 2024, though easy to miss as a passing reference in a long video.
In some senses I really couldn’t blame them for spreading out donations with that kind of windfall. I’d personally just donate the maximum amount per year that I could still write off on taxes which is 60% of annual money income or 30% equivalent income if doing direct stock transfers.