#292 — How Much Does the Future Matter?

Episode Summary

Executive Summary: Sam Harris opens with brief commentary on the Trump FBI search and the attack on Salman Rushdie before pivoting to a wide-ranging interview with Will MacAskill about effective altruism, longtermism, existential risk, value lock-in, and AI. The conversation argues that future people matter morally, that current institutions badly underweight them, and that technology is advancing faster than wisdom, making this a pivotal era for safeguarding civilization.

Main Topics: Opening commentary on current events and free speech: Harris discusses the Mar-a-Lago FBI search and the attack on Salman Rushdie, warning against politicized institutions and criticizing secular and religious cowardice toward Islamist intimidation. Effective altruism and its core premise: MacAskill defines effective altruism as using evidence and reason to do the most good possible with time and money, emphasizing that markets, politics, and common sense often neglect the most important beneficiaries. Longtermism and concern for future people: The discussion argues that future generations can be morally significant even if they do not yet exist, and that present actions can strongly shape whether humanity thrives, stagnates, or collapses. Pascal’s mugging, expected value, and decision theory: Harris raises concerns about tiny probabilities multiplied by enormous future value. MacAskill rejects extreme fanatical conclusions, arguing that longtermism should not rest on absurdly small probabilities and that bounded value intuitions matter. Hinge moment in history and technological acceleration: Both speakers agree that the present is unusually consequential because technological progress is rapid, compounding, and potentially far ahead of wisdom, giving current generations outsized influence over the long-term future. Value lock-in, moral progress, and political incentives: MacAskill warns that a narrow ideology could become globally dominant and persist for centuries or longer, but also notes a paradox: preserving conditions for moral progress may require locking in liberal norms such as free debate and anti-authoritarian constraints. Artificial intelligence as an existential and governance challenge: They discuss AI as the clearest example of a technology that could disempower humanity or enable catastrophe. MacAskill argues for prioritizing safety, governance, and safer AI systems before more agentic or dangerous ones.

Key Arguments: Doing good effectively requires evidence-based prioritization, not just responding to the most emotionally salient causes. Future generations count morally, and present policy choices can determine whether the long-term future is flourishing or catastrophic. The future is not abstract in a morally irrelevant way: if civilization survives, future people are guaranteed to exist, making today’s choices consequential. Pascal-style reasoning can become absurd if applied to tiny probabilities and astronomical value; concern for the future should not rely on fanatic expected-value arguments. Human psychology is poorly adapted to large-scale, distant, and future-oriented moral reasoning, so communities and institutions must compensate for that bias. The current era may be a hinge moment because technological growth is unusually fast compared with both the past and likely future. Political systems are structurally short-termist, so institutions and culture must be redesigned to represent future generations. Value lock-in is a real risk: a bad ideology, empowered by technology, could become globally dominant and freeze moral progress. AI safety must be integrated into AI development itself, not treated as an afterthought like 'not falling down' for bridges. A safer path is to accelerate beneficial, non-agentic AI tools while slowing or carefully governing more dangerous, goal-driven systems.

Data Points: FBI search / Mar-a-Lago raid: No numeric value stated - Harris discusses the FBI search of Trump’s Mar-a-Lago property and cautions against drawing conclusions before facts emerge. Rushdie attack severity: On life support - Harris says Salman Rushdie was attacked on stage and reports he sounded to be on life support. EA fundraising: Over a billion dollars - MacAskill notes the Center for Effective Altruism has raised over a billion dollars for charities. 100 Days of Giving: $10,000 a day for 100 days - Harris describes a Waking Up campaign donating $10,000 daily to vetted charities. Global births during conversation: 3,000 to 5,000 people - Harris estimates that number of people are born worldwide while they speak. Humanity’s remaining habitable time: 500 million to 1.3 billion years - Harris references the Sun’s long-term behavior as a basis for future planning. Economic growth rate: About 2% to 3% per year - MacAskill uses current growth rates to argue that civilization is moving very quickly through technological development. Doubling time of the economy: About 20 to 25 years - Harris and MacAskill mention compounding growth in modern economies. Agricultural-era growth: About 0.1% per year - MacAskill contrasts modern growth with the much slower pace in farming societies. AI survey median timeline: 37 years - MacAskill cites a survey of machine learning researchers on when human-level machine intelligence might arrive. AI existential risk estimate: 5% - He says researchers gave a typical probability of an extremely bad AI outcome, including extinction, of about 5%. Implied AI lifetime harm risk: 2.5% - Harris combines the 50-50 human-level AI timeline with the 5% catastrophe estimate to frame lifetime risk. Current AI math benchmark: About 50% accuracy - MacAskill says a recent model can solve early college-level math proofs with around half success.

Pivotal Quotes: "how can we do as much good as possible with our time and with our money" — Will MacAskill: MacAskill defines effective altruism at the start of the interview. "the problem has always been larger than mere Islamic extremism and its intolerance to free speech" — Sam Harris: Harris explains his view that the Rushdie fatwa was enabled by cowardice and confusion among secular liberals, not just religious extremists. "we are moving much faster through that technology to the E than for almost all of human history" — Will MacAskill: MacAskill argues that the present is an unusually consequential technological era.

Implications: Listeners are urged to think less sentimentally and more strategically about moral action, especially around AI, pandemics, and institutions that affect future generations. The episode frames long-term stewardship as a present-day ethical duty with potentially civilization-scale consequences.