Erik Hoel on Effective Altruism, Utilitarianism, and the Repugnant Conclusion

Episode Summary

Executive Summary: Russ Roberts and neuroscientist Eric Hoel examine effective altruism’s strengths—its insistence on measurable impact and willingness to fund neglected causes—but argue its utilitarian core pushes it toward repugnant conclusions. They critique thought experiments like the trolley problem and shallow pond for scaling poorly, question commensurability across lives, and warn that EA’s influence can shape harmful funding priorities, especially in AI safety, if humility and qualitative moral limits are absent.

Main Topics: What Effective Altruism Gets Right (Priority: 5/5): EA is framed as a 'Moneyball' approach to charity: use evidence and logic to direct dollars where they do the most good rather than where donors feel best. Both speakers agree this is an important corrective to sentimental or low-impact giving. Utilitarianism and Its Thought Experiments (Priority: 5/5): Hoel argues EA’s philosophical roots in utilitarianism depend on trolley-problem and shallow-pond intuitions that may work in simple cases but break down at scale, leading to morally troubling conclusions. The Repugnant Conclusion and Moral Commensurability (Priority: 5/5): They discuss Parfit’s repugnant conclusion: maximizing total utility can imply a world with many miserable people is preferable to fewer flourishing ones. This raises the deeper issue of whether well-being can be aggregated across people at all. AI Safety as a Practical Test Case (Priority: 4/5): Hoel argues EA funding has real-world consequences in AI safety, where utilitarian sympathy toward digital minds may bias the movement toward 'enslaving' AIs rather than regulating or banning dangerous systems. Qualitative Morality, Human Nature, and Religion (Priority: 4/5): Hoel contends utilitarianism ignores qualitative values like dignity, nature, humanness, and sanctity of human life—values often grounded in religion but also in ancient philosophy and cultural inheritance. Economics, Free Trade, and Utilitarian Reasoning (Priority: 4/5): Roberts connects the critique to economics, especially cost-benefit logic in trade policy, arguing that comparing dollars or welfare across people can miss dignity, local knowledge, and irreversible losses. EA’s Useful Expansion of Charity (Priority: 3/5): Despite criticism, Hoel praises EA for expanding the set of causes that can count as charity, including online writing, asteroid mining, and other unconventional long-horizon or science-fiction-like projects.

Key Arguments: EA usefully pushes donors to think about actual impact rather than emotional satisfaction, but it becomes distorted when it treats morality as a maximization problem. The trolley problem and shallow pond are persuasive in isolation, but they do not scale cleanly to real-world decisions involving institutions, uncertainty, and rights. Utilitarianism encourages arbitrage across persons and causes, but that logic can lead to the repugnant conclusion: more people living worse lives can be deemed better overall. Morality is not fully commensurable; good and evil are not simple 'mounds of dirt' that can be traded off without remainder. EA’s practical funding priorities, especially in AI safety, show that philosophical assumptions can affect real institutional choices. A strong alternative is humility: give fixed shares of income, measure local impact when possible, and avoid overconfidence in abstract welfare calculations. Qualitative goods such as dignity, human nature, nature itself, and sanctity cannot be captured well by utilitarian sums. Even if EA is flawed philosophically, it has still done substantial good by broadening what people see as worth funding.

Data Points: Podcast date: August 31st, 2022 - Introductory metadata for the EconTalk episode Years of EconTalk archives: Since 2006 - Roberts notes the show’s archive depth Seeing-eye dog cost example: $40,000 - Used as a contrast with lower-cost global interventions Blindness prevention example: 500 cases - Approximate number of blindness cases that same $40,000 might prevent in the third world Shallow pond analogy source: Peter Singer, Famine, Affluence, and Morality - Referenced as a foundational EA thought experiment Repugnant conclusion scale example: Many people at borderline subsistence - Used to describe the outcome of maximizing total utility EA funding scale: Billions - Describes the scale of money influencing the movement AI safety funding concentration: A very small handful of companies - Only a few firms are developing advanced AI systems with millions in training costs Coal reserve idea: Leave some coal in the ground - McCaskill example of an unconventional long-termist charity cause FTX foundation reference: Head of this foundation - Mentioned in connection with academic work on long-run expected value and risk

Pivotal Quotes: "Effective altruism is the money ball of charity." — Eric Hoel: Hoel’s concise definition of EA’s appeal and method "What ends up happening is that everyone then ends up with a life that's only sort of just barely above subsistence level because everyone's been all the extra happiness has been arbitraged away such that it's been sort of fairly distributed among everybody else." — Eric Hoel: Critique of utilitarian maximization and the repugnant conclusion "Morality is not a market." — Eric Hoel: Hoel’s central objection to treating well-being as tradable across persons

Implications: Listeners are left with a tempered view: EA can improve philanthropy, but if it keeps treating morality as pure optimization, it risks endorsing harmful tradeoffs. Future debates over AI, charity, and policy will likely hinge on whether welfare can be measured and aggregated at all.