I am open to work.
Vasco Grilošø
Karma: 8,531
Thanks, Laura!
Hi Hannah and William,
Have you considered doing a similar analysis for farmed insects?
Thanks for the update! Which fraction of the marginal donations to AWF go to invertebrate or wild animal welfare?
Thanks, Carolina! Just one note, you do not have to tag me, as I receive email notifications even if you do not.
If I understand correctly, of your spending of 248 k$ on Nourishing Tomorrow in 2023, 80 k$ was unrestricted, and 168 k$ (= (248 ā 80)*10^3) was restricted to that program. I think you are saying the program as a whole caused 100 k$ of additional unrestricted donations that year. If so, the program caused 0.403 $ (= 100*10^3/ā(248*10^3)) of additional unrestricted donations per $ spent (in reality, it is lower due to overhead[1]). That is less than 1 $, and I think the multiplier for the unrestricted funds is even lower due to diminishing returns[2], so it looks like you should not be spending unrestricted funds on the program. Am I missing something?
Break-even analĀyĀsis of D3, long-chain omega-3, and mulĀtiĀviĀtamin-mulĀtiĀminĀeral supplementation
Regarding prioritization: You can find details on how we allocate funding across programmatic areas in our financial statements.
I have checked your financial statements for 2023. It would be great if you added how much restricted and unrestricted funds you spend on each program.
I donāt follow this comment. Youāre saying Vasco gives you X now, 2X to be paid back after k years. You plan to spend X/ā2 now, and lock up X/ā2, but somehow borrow 3/ā(2X) money now, such that you can pay the full amount back in k years?
Nitpick. (3/ā2) X, not 3/ā(2 X).
If one expects investments to grow more (in real terms) than the product ācost-effectiveness of altruistic spending conditional on survivalā*āprobability of survivalā will decrease, it makes sense to invest as much as possible now, and then donate as much as possible later. Funders of altruistic interventions should try to equalise the product I just mentioned across years (otherwise, they should move their spending from the worst to the best years).
Nice points, Sasha!
On the 1st point, I think one should borrow money from banks until the conditions for borrowing additional money become as good as those of the available bets, and then get money both ways afterwards. Refusing a bet which is beneficial relative to nothing because there are loans with better conditions suggests one should be asking for more loans.
On the 2nd point, I wondered about the possibility of Greg not fulfilling the bet in order to decrease AI risk further, but I believe the world will look roughly the same way as now in terms of risk. So I expect Greg will not be much more worried than now, and therefore will fulfill the bet.
Hi @Laura Duffy,
In the quantitative model, you calculated the ratio between values for humans and animals. Should you have calculated the ratio between values for animals and humans, considering you are estimating the welfare range of animals relative to that of humans?
Hi Chris,
I estimate broilers in conventional and reformed scenarios have a welfare of ā2.27 and ā0.161 AQALY/ābroiler-year, and hens in conventional cages and cage-free aviaries have a welfare of ā1.69 and ā0.333 AQALY/āhen-year[1]. I estimate shrimps on ongrowing farms not stunned, stunned with ice slurry (although note this is rarely done properly), and eletrically stunned before slaughter have a welfare of ā8.77, ā4.40 and ā4.19 AQALY/āshrimp-year. For comparing effects across species, I get estimates for the welfare in QALY/āanimal-year multiplying the welfare in AQALY/āanimal-year by Rethink Prioritiesā (RPās) median welfare ranges based on my guess that the welfare per animal-year of fully happy life is proportional to the welfare range.
To calculate the welfare in AQALY/āanimal-year, I assumed the welfare of annoying, hurtful, disabling, and excruciating pain are ā0.1, ā1, ā10, and ā100 k AQALY/āpain-year, which imply each of the following neutralise 1 day of fully happy life:
10 days (= 1ā0.1) of annoying pain.
1 day of hurtful pain.
2.40 h (= 24ā10) of disabling pain.
0.864 s (= 24*60^2/ā(100*10^3)) of excruciating pain.
There is lots of uncertainty. Feel free to make copies of the sheets I linked above, and update the pain intensities in the tab āPain intensitiesā.
- ^
To illustrate, 2.27 fully happy broiler-years, as in healthy broilers in a sanctuary, are needed to neutralise 1 broiler-year in a conventional scenario, which has a welfare of ā2.27 animal quality-adjusted life-years (AQALYs) per broiler-year.
Thanks for the post! I strongly upvoted it.
Are you planning to apply the framework to:
Global health and development interventions? Decreasing the risk of human extinction corresponds to saving human lives in worlds where human population would get to 0 without the intervention. If this changes the longterm value of the future, then I do not see why saving human lives for a larger population would not.
Animal welfare interventions? I estimate the room to improve wild animal welfare is 10.5 M (= 3.03*10^16/ā(2.88*10^9)) times as large as that to improve human welfare. I think this suggests increasing wild animal welfare by 1 % would have a much greater effect on the longterm value of the future than increasing human welfare by 1 %.
How do you suggest decision-makers decide on the parameters of the model? It is very difficult to predict specifics of the future over a few decades from now, so I do not know how one can make informed choices about effects on the longterm value of the future.
You assume that decreasing existential risk over a given period does not affect the existential risk (or welfare) after that period, right? One could spend resources to decrease existential risk in 2025, but this will imply having (counterfactually) less resources in 2026. So a decreased risk in 2025 is partially offset by an increased risk in 2026.
I think expected future earnings, including salaries and appreciation of equity, would go down in most cases. I thought you would agree because you said āeven though it will be hard to give up the moneyā.
Hi Manuel,
You may like this post by nostalgebraist. I am open to betting up to 10 k$ against short AI timelines. I understand this does not work for people who think doom or utopia are certain soon after AGI, but I would say this is a super extreme view. Banks may offer loans with better conditions, but, as long as my bet is beneficial, one should take the bank loans until they are marginally neutral, and then also take my bet.
Thanks for sharing! Are you planning to account for differences in cost-effectiveness across areas?
Thanks for the post, Holly. Strongly upvoted. I did not find the post that valuable per se, but it generated some good discussion.
If you care about protecting the world, you will quit, even though it will be hard to give up the money and the prestige and the hope that they would fix the problem.
People at leading AI companies can earn hundreds of thousand of dollars per year, so quitting could plausibly decrease their donations by 100 k$/āyear. I estimate donating this to the Shrimp Welfare Project (SWP) would decrease as much pain per year as that needed to neutralise the happiness of 1.25 M human lives (= 100*10^3*639/ā51). Do you think the benefits of quitting outweight this? I do not, so I encourage people at leading AI companies to simply donate more to SWP. I imagine no one would quit if there were actual human lives on the line (instead of shrimp which are not helped).
Thanks for the post, Saulius!
Your estimates imply the Shrimp Welfare Projectās (SWP) Humane Slaughter Initiative has been 38.6 (= 25*10^3/ā648) times as cost-effective as the future work of Animal International Poland. This suggests someone working here who is 10 % more cost-effective than the 2nd best candidate for their role, and donates 10 % more of their gross salary to SWP has 38.6 times as much impact through donations than through work. This illustrates why I think donating more and better is the best strategy to maximise impact for the vast majority of people working in impact-focussed organisations (such as Animal International Poland).
For your assumption that disabling pain is 60.0 (= 1/ā0.01667) times as intense as hurtful pain, and my guesses that this is as intense as fully health life, and that fully healthy life in chickens is 33.2 % as intense as in humans (given Rethink Prioritiesā (RPās) median welfare range of chickens of 0.332), 1 day of disabling pain in chickens (1 DCDE) is as bad as 0.0545 DALYs (= 60.0*0.332/ā365.25). So I would say your estimate for the future cost-effectiveness of Anima International Poland is, based on your own pain intensities, equivalent to averting 35.3 DALYs per $ (= 648*0.0545), 3.55 k (= 35.3/ā0.00994) times as cost-effective as GiveWellās top charities (neglecting their effects on animals). I estimated cage-free campaigns are 462 times as cost-effective as GiveWellās top charities.
Thanks for the great clarifications, Daniela!
It is interesting RPās work on wild animal welfare is not supported by unrestricted funds. It suggests the people at RP responsible for allocating the unrestricted funds think there is other work from RP which is more cost-effective at the margin. How are unrestricted funds allocated? I think it would be great for RP to be transparent about this considering donations become unrestricted funds by default.
Will donations restricted to RPās work on invertebrate welfare (including farmed, wild, and other invertebrates) also not go towards vertebrate welfare (including humans, and vertebrate animals)? Which fraction of the funds supporting invertebrate welfare are unrestricted? I asked these questions about wild animal welfare, but I am actually specially interested in invertebrate welfare.