Episode Summary
Executive Summary: Will MacAskill argues that long-termism is the view that positively influencing the far future is a key moral priority. The conversation covers why future people matter, why civilization may last vast spans of time, how current actions could shape that trajectory, and which interventions seem promising, especially pandemic prevention, AI safety, clean energy, and preserving liberal-democratic values.
Main Topics: What long-termism is and why it matters: Long-termism is framed as prioritizing the far future because future people matter morally and there may be vastly many of them. MacAskill distinguishes the weak claim that it is a key priority from stronger claims that it is the overriding priority. The scale of the future and moral weight of future people: The discussion emphasizes that humanity could persist for hundreds of thousands to trillions of years, making the number of future lives enormous relative to present ones. This creates a strong moral case for caring about long-run outcomes. How we can affect the future: MacAskill argues that we can influence the far future by reducing extinction/catastrophe risk and by shaping the values and institutions that guide future civilization, including preserving information and robust social norms. Concrete priority areas: pandemics and AI: The conversation highlights worst-case pandemics, synthetic biology, wastewater screening, far-UVC sterilization, and AI safety/alignment as major long-termist cause areas with potentially pivotal effects on civilization’s trajectory. Contingency of values and the risk of lock-in: MacAskill argues that present-day values are historically contingent, citing slavery, religion, monogamy, and cultural norms, which implies future values could also be locked in by today’s actions—especially through AI or geopolitical shifts. Objections: tractability, fanaticism, and demandingness: The interview addresses worries that long-termist interventions lack feedback loops, encourage fanatical low-probability reasoning, or become morally over-demanding. MacAskill responds that these issues affect many moral choices and that current efforts are still far from extreme sacrifice. Stagnation and long-run civilization risk: A substantial section explores the possibility that economic and technological progress may stagnate as ideas get harder to find and population growth plateaus, raising concerns about prolonged periods of elevated existential risk and value lock-in.
Key Arguments: Future people matter morally, and there may be vastly more future people than present people, so the future carries enormous moral significance. Even if one is uncertain about population ethics, long-termism still follows from caring about the quality of future lives, not just whether future lives exist. We can affect the far future through extinction-risk reduction, institutional design, preserving knowledge, and shaping values before they become locked in. Pandemic prevention and AI safety are especially promising because they can avert catastrophic or civilization-ending outcomes and influence the long-run trajectory. Values are historically contingent; therefore, the moral beliefs that dominate the future may depend heavily on actions taken now. Long-termism is not a demand to sacrifice everything now; practical spending on the far future is still tiny relative to total societal resources. Many criticisms of long-termism, such as lack of feedback loops or fanaticism, apply broadly to moral decision-making, not just to long-termist causes. A stagnating civilization could remain in a prolonged period of high risk, making progress toward safer and better institutions especially urgent. The future could be shaped not only by human survival but also by what kind of civilization survives—e.g., liberal democracy versus authoritarianism or misaligned AI systems. Robustly good actions—like cleaner energy, lower war risk, and movement-building—can be worthwhile across many moral views, not just long-termism.
Data Points: Humanity’s age: about 300,000 years - Used to compare humanity’s current age with possible future durations. Low-end future ratio: about 1,000 future people per person alive today - If humanity lasts as long as a typical mammal species. Possible future duration: hundreds of millions to hundreds of trillions of years - Depending on whether civilization lasts until Earth is uninhabitable or becomes interstellar. Current share of societal resources for future generations: around 0.1% - MacAskill suggests society currently spends very little on safeguarding the far future. Pandemic deaths from COVID-19: tens of millions - Used as a reference point for what much worse engineered pandemics could exceed. FAR-UVC: a narrow spectrum of light - Proposed as a physical method that could sterilize rooms/air and reduce pathogen spread. AI takeover risk (MacAskill estimate): about 3% - His rough estimate of misaligned AI takeover risk in the book discussion. Chance of dying of cancer: about 20% - Compared with AI risk to calibrate uncertainty about personal mortality. Engineered pandemic extinction risk this century: about 0.5% - His rough estimate for extinction from engineered pandemics. Great-power war in our lifetimes: about 40% - MacAskill’s estimate of major war risk. Seatbelt comparison: 1 in 300 million - Used to illustrate that people routinely act on very low-probability, high-stakes risks. Lead exposure intervention: lead lowers IQ and increases violence - Explained in relation to the LEAP donation and broader long-term social effects. Fossil-fuel particulates: about 3.5 million deaths/year - Cited as a near-term benefit of clean energy transitions. EU life expectancy loss from particulates: about 1 year - Example of health gains from reducing fossil-fuel pollution. Coal reserve example: about 300 years of coal at current consumption - Used to illustrate why keeping coal in the ground could matter after catastrophe and recovery. Population projection: plateau around 11 billion - Used in the stagnation argument about limits to growth through population expansion. Factory farm focus: tens to hundreds of millions of bednets; thousands of lives saved - Cited as a success of effective altruism’s more concrete, near-term work. Survey timing: deadline extended by two days - Housekeeping notice about 80,000 Hours user survey. Marketer role salary: around £60,000 - Housekeeping notice about hiring for 80,000 Hours.
Pivotal Quotes: "Long-termism is the view that positively influencing the long-term future is a key moral priority of our time." — Will MacAskill: Definition of long-termism early in the interview. "Harm is harm wherever it occurs." — Will MacAskill: Used to justify caring about people far in the future as much as those near in time. "There’s this strong asymmetry where I think there’s an enormous number of ways the future could go, such that we lose almost all of our potential." — Will MacAskill: Explaining why the future seems more likely to be mediocre or bad than maximally good.
Implications: For listeners, the episode argues for treating long-run risks and value-shaping work as serious priorities, not abstract philosophy. For the field, it supports funding and talent toward biosecurity, AI safety, liberal institutions, and robustly good interventions.