#467 — EA, AI, and the End of Work

Episode Summary

Executive Summary: Sam Harris and Will MacAskill revisit effective altruism on the 10th anniversary of Doing Good Better, discussing EA’s recovery after the FTX/Sam Bankman-Fried scandal, the growth of giving, and the movement’s core cause areas: global health, animal welfare, pandemic preparedness, and AI risk. They also debate whether EA should broaden to include harder-to-quantify but potentially massive opportunity-cost issues and positive-flourishing goals.

Main Topics: EA’s recovery after FTX and movement growth (Priority: 5/5): MacAskill argues that despite the reputational hit from FTX and Sam Bankman-Fried, EA’s underlying ideas have regained momentum, with growth in donations, pledges, and conference participation. Global health and development as a flagship EA cause (Priority: 5/5): The conversation emphasizes that high-quality evidence shows some aid interventions save lives at very low cost, making global health one of EA’s most established and effective focus areas. Animal welfare and factory farming (Priority: 4/5): MacAskill defends farm animal welfare as a major EA priority, citing the scale of suffering in factory farming and the success of corporate cage-free campaigns. Pandemic preparedness and biosecurity (Priority: 4/5): The speakers discuss how pandemic prevention remains underfunded despite COVID-19, with practical interventions like wastewater monitoring, air sterilization, and lab safety seen as high-leverage. AI risk and accelerating capabilities (Priority: 5/5): MacAskill says AI progress has surprised experts and is advancing on a stable exponential curve, making AI safety and the automation of AI research a central concern. Expanding EA beyond negative suffering reduction (Priority: 4/5): Harris presses MacAskill on whether EA should also account for opportunity costs and positive human flourishing, not just preventing harm, and whether the movement should broaden its moral scope. Moral weirdness and long-run ethical progress (Priority: 4/5): MacAskill defends taking unusual moral ideas seriously—like concern for animals or digital minds—arguing that history shows many once-ridiculed views later became moral common sense.

Key Arguments: EA’s influence has rebounded after FTX; the movement’s core ideas remain persuasive and continue to attract donors and participants. Global health interventions are among the most cost-effective ways to save lives, with strong evidence from randomized trials and major life-saving impact. Factory farming causes immense suffering at massive scale, and relatively modest funding has already improved conditions for billions of chickens. Pandemic preparedness is a neglected but practical area where relatively straightforward investments could reduce both natural and engineered outbreak risks. AI capabilities are advancing faster than many experts expected, and the point where AI accelerates AI research itself may trigger a major capability jump. EA should not be limited to conventional or socially comfortable causes; it should remain open to morally serious but initially strange ideas. Harris argues that EA may underweight opportunity costs and positive flourishing, not just suffering reduction, especially in politics and social systems. MacAskill agrees that future generations may judge current society as missing out on major goods, and that long-run ethics should include both reducing suffering and enabling better lives.

Data Points: 10% pledges from podcast listeners: 1,200 - MacAskill says listeners of this podcast have made 1,200 pledges to donate 10% of income. Donations moved via podcast influence: Over $30 million - MacAskill estimates donations redirected through the podcast audience’s pledges. Lives saved via GiveWell donations: Over 340,000 - MacAskill says GiveWell-directed donations have likely saved this many lives. Cost per life saved in global health: About $5,000 per life - Used to compare global health interventions with U.S. life-extension costs. Cost to give one year of life in the U.S.: About $50,000 - MacAskill contrasts U.S. healthcare cost-effectiveness with global health aid. Farm animals raised and slaughtered annually: About 90 billion - MacAskill cites the scale of factory farming worldwide. Chickens improved by cage-free campaigns in the U.S.: 3 billion per year - He says these chickens now live in somewhat better conditions due to corporate pledges. Fulfillment rate of cage-free pledges: 92% - MacAskill reports most corporate cage-free commitments were fulfilled. Funding for cage-free campaigns: Tens of millions of dollars - He emphasizes the relatively low cost of achieving large animal-welfare gains. Projected deaths from USAID dismantling: 14 million - Harris cites a Lancet study estimating deaths over five years from infectious disease. Projected deaths under age five from USAID dismantling: 4.5 million - Harris highlights the child mortality component of the Lancet estimate. Alternative discounted estimate of USAID deaths: 1.4 million total; 450,000 under age five - Harris suggests even a tenfold discount still implies catastrophic harm. Pandemic risk estimate: One in three - MacAskill says he thinks there is roughly a one-in-three chance of waves of new pandemics from lab leaks or related risks. Lab leak frequency in high-security labs: About 1 virus leak per 100 person-years - MacAskill uses this to argue containment is difficult. AI software-engineering capability growth: Doubles every 4–6 months - MacAskill describes current AI progress as a stable exponential trend. AI research forecast surprise: Only 2%–5% expected math Olympiad-level performance by 2025 - Harris references a forecast showing experts underestimated AI progress.

Pivotal Quotes: "the underlying ideas are very good" — Will MacAskill: MacAskill explains why EA has recovered despite the FTX scandal. "I think factory farming is one of the worst atrocities that humanity is committing today." — Will MacAskill: He defends animal welfare as a major EA priority. "we should be trying to think, yeah, not merely just how can we eliminate obvious causes of suffering, but actually how can we perhaps have a life that's, you know, radically better today than today" — Will MacAskill: MacAskill responds to Harris’s point about positive flourishing and opportunity costs.

Implications: EA is broadening from a narrow focus on measurable suffering to include long-run risks, moral uncertainty, and human flourishing. For listeners, the takeaway is to think in terms of scale, evidence, and opportunity cost when choosing causes.