I am looking for work, and welcome suggestions for posts.
I am looking for work. I welcome suggestions for posts. You can give me feedback here (anonymous or not). Feel free to share your thoughts on the value (or lack thereof) of my posts.
I can help with career advice, prioritisation, and quantitative analyses.
Thanks for the comment, Michael!
I wonder how things would change if you had short AI timelines or otherwise assumed the impacts would be cut off after 5 or 10 years, say.
I may think about this tomorrow. I like Ege Erdil's median time until full automation of remote work of 20 years.
Is the cost-effectiveness based on increases in life expectancies only, or also improvements in quality of life? Increases in life expectancy would be the main drivers for effects on animals, and quality of life changes would probably have minimal effects on animals.
The cost-effectiveness estimates from CEARCH, and Joel's guess account for effects on mortality, morbidity, and consumption. I have now clarified the following. "I assume donating to HIPF from CEARCH increases human-years 12 times as cost-effectively as GiveWellâs top charities, as I estimate the cost-effectiveness of donating to HIPF accounting only for [mortality, morbidity, and economic effects on] humans is 12 times that of GiveWellâs top charities". I have also updated my estimate of the increase in the living time of humans caused by GiveWell's top charities. "I estimate GiveWellâs top charities increase the living time of humans by 0.0157 human-year/$, which is the ratio between the period life expectancy at birth in low income countries in 2023 of 64.9 human-year/life, and the mean cost of saving a life donating to those charities in 2021 to 2023 of 4.13 k$/life". Previously, I was assuming a period healthy life expectancy at birth of 51 years, as, according to Open Philanthropy (OP), âGiveWell uses moral weights for child deaths that would be consistent with assuming 51 years of foregone life in the DALY framework (though that is not how they reach the conclusion)â.
Thanks for clarifying, Michael.
(I'm assuming we're ruling out an average welfare of exactly 0 or assigning that negligible probability.)
I would agree any particular value for the welfare per animal-year has a negligible probability because my probability distribution is practically continuous, such that there is lots of values around any particular one.
Thanks, Michael. Would the approaches you mention only recommend acting on the basis that wild animals have negative lives if the probability of this was sufficiently high? If so, why would my estimates of the probability of soil nematodes, mites, and springtails having negative lives of 58.7 %, 55.8 %, and 55.0 % be too low, but, for example, 70 % be sufficiently high?
Thanks for the comment, Toby.
Outsourcing the welfare estimates to Gemini seems like a risky move to me. It's a key part of the whole analysis, but is an extremely challenging question to begin answering. What's the reason to expect Gemini to be able to do a good job of this, given the blind spots we know current AI models still have?
I put little trust in Gemini's estimates. However, my sense is that most people working on wild animal welfare would guess soil nematodes, mites, and springtails have negative lives. I have now clarified this in the post. In addition, Gemini's estimates are in close agreement with Ambitious Impact's estimate for wild bugs.
Gemini provided best guesses for soil nematodes, mites, and springtails of -67 %, -44 %, and -38 %, which are 1.60, 1.05, and 0.905 times Ambitious Impactâs estimate of -42 % for wild bugs based on their deprecated welfare points system.
If I'm understanding right, this would flip all your conclusions on their head, and instead of trying to eliminate wild animal habitats, the top priority would be to increase them?
Right.
Such extreme sensitivity to highly uncertain quantities strikes me as a strong reductio ad absurdum argument against this approach to decision making on this kind of question. Otherwise we find ourselves oscillating wildly between "destroy all nature" and "destroy all humans" on the basis of each piece of new information, never being especially confident in either.
Uncertainty about whether wild animals have positive or negative lives only directly translates into uncertainty about whether one should increase or decrease wild-animal-years at the margin, which is not absurd, and neither is my recommendation of saving human lives cost-effectively. Killing all wild animals or humans are not live options.
I like Bayesianism and expected value maximization as a framework for decision making under uncertainty, but when considering situations with enormous amounts of value described by extremely speculative probability estimates, I think we probably need to approach things differently (or at least adapt our priors so as to be less sensitive to these kind of problems). Something like Holden Karnofsky's approach here (which Anthony DiGiovanni shared with me on a recent post on insect suffering).
Approaches neglecting the effects on wild animals would be implicitly considering them negligible. For this to be the case, I think one would need an unreasonably certain prior that wild animals have welfare almost exactly equal to 0.
I plan to recommend people donate to CEARCH's High Impact Philanthropy Fund (HIPF) in a post I am writing which I will share in this thread once it is published.
I have now published the post where I recommend HIPF. It looks into the cost-effectiveness of interventions accounting for soil nematodes, mites, and springtails.
Btw, I think itâs unlikely that nematodes are sentient because they are so simple. The most commonly studied one has like 300 neurons. But I see they are excluded from your estimate anyway because they are not arthropods.
I think nematodes matter. I calculate soil nematodes, mites, and springtails have (in expectation) a welfare of -4.36*10^-6, -1.57*10^-5, and -2.35*10^-5 QALY/animal-year, and an annual welfare of -296 k, -13.9 k, and -10.4 k times that of humans.
I plan to publish a post about this on Tuesday.
I have now published a post about the cost-effectiveness of interventions accounting for soil nematodes, mites, and springtails.
Right. 93.1 % of the increase in the welfare of soil animals resulting from increasing cropland comes from decreasing nematode-years. For a welfare per nematode-year 10 % as high, the cost-effectiveness accounting for target beneficiaries and soil animals of donating to HIPF would become 3.28 times the past cost-effectiveness of HSI[1], which is 16.2 % (= 3.28/20.3) of the ratio I present in the post of 20.3.
I got this updating this cell to a value 10 % as high.