Cryonics with a 3% chance of revival yields about as many years of life per dollar as GiveWell’s top charity programs, AI safety researcher Max Harms calculates
Cryonics with a 3% chance of revival yields about as many years of life per dollar as GiveWell’s top charity programs, AI safety researcher Max Harms calculates
On November 16, the Effective Altruism Forum published the essay “Cryonics as a Cause Area”. Effective altruism is a movement that directs donations according to the benefit they provide per dollar. Harms proposes including cryonics, the preservation of people after death in the hope of reviving them, in charitable giving portfolios. Nectome, a company that sells this preservation service, hired him to write about cryonics; the essay ends by promoting its discount cards.
Harms measures benefit in quality-adjusted life years (QALY). For GiveWell’s top programs, he uses a standard estimate of 0,01 QALY per dollar, or 100 dollars per year of life; GiveWell itself estimates that these programs save a life for 3500–5500 dollars.
Harms calculates how long a revived person would live in a world that has developed the technology to revive people and overcome aging. In that world, people die only from accidents, violence, and a few other rare causes. He takes the annual risk of death from statistics for wealthy countries, excluding crashes and poisonings, which he believes will largely disappear, and adds his own estimate for other causes. The result is 0,005% per year, implying an average lifespan of 20 thousand years.
Assuming no global catastrophe, Harms puts the chance of reviving someone preserved today at a pessimistic 1/30, or about 3%. Multiplying that probability by 20 thousand years and a quality-of-life factor of 1,5 (future life would be 1,5 times as good as a healthy life today), he arrives at about 1000 QALY per person preserved. GiveWell’s top programs provide the same benefit for 100 thousand dollars; preservation currently costs 29–250 thousand dollars.
Harms acknowledges that he is comparing “apples and oranges”: GiveWell’s estimates assume that children whose lives are saved will live ordinary lifespans. If AI also radically extends life in poor countries, saving a life for 4 thousand dollars would outperform preservation costing tens of thousands.
According to Harms, preservation becomes cheaper at scale. Part of today’s price is a one-time payment whose investment returns fund indefinite storage: with annual expenses of 2200 dollars and returns of 4% above inflation, that payment is 55 thousand dollars. Nectome plans to store clients in industrial freezers without liquid nitrogen. Harms calculates that electricity in a large storage facility would cost no more than 10 dollars per person per year, reducing the payment to a couple of hundred dollars.
If millions of people were preserved each year (Harms estimates that fewer than 50 are preserved annually today), he estimates a price of 15 thousand dollars, mostly for specialist labor and logistics. If the technology and institutions mature and the chance of revival reaches 50%, his calculation yields 0,6 years of life per dollar, 60 times the 0,01 provided by GiveWell’s top programs.
Discounting, which assigns less value to benefits received further in the future, would reduce the estimated value of cryonics to about 0,5% of its undiscounted value. At a rate typical of human preferences, extending a lifespan from 80 to 20 thousand years provides only 32% more benefit than saving an infant who lives to 80 (without discounting, the benefit is 250 times as large). Harms objects that this rate produces preferences that change over time: young people accept dying at 80, but at 70 they want to live longer.
Harms leaves out of his calculation an argument made by Nectome co-founder Aurelia Song: people accelerating AI development to overcome death might slow down if preservation gave them a fallback. To those prepared to devote all their efforts to addressing AI risks, he writes:
“About 60 million people will die next year, roughly two souls every second. We need to make room in our hearts and our portfolios to try to save them too, alongside the world as a whole.”