Episode Summary
Executive Summary: Holden Karnofsky explains why Open Philanthropy is increasingly split between global health and longtermism, and lays out his case that we may live in the most important century ever. He argues AI-enabled scientific acceleration, digital people, and rapid growth could compress millennia of change into decades, making this era unusually consequential and demanding urgent, serious attention.
Main Topics: Open Philanthropy’s strategic split: Holden describes promoting Alexander Berger to co-CEO so global health/well-being and longtermist work can be run as distinct missions with different methods, bottlenecks, and scales. Why this could be the most important century: The core of the conversation is Holden’s argument that humanity may be at a uniquely pivotal point because AI and other technologies could determine the trajectory of a galaxy-spanning future. AI as a driver of explosive progress: He discusses PASTA—processes that automate scientific and technological advancement—and why AI could accelerate innovation, economic growth, and scientific discovery much faster than most people expect. Digital people and the future of society: Holden makes the case that mind uploads or digital persons may eventually be feasible, would likely count morally, and would radically transform labor, identity, rights, and the structure of civilization. Economic growth, exponentials, and timing: A major thread is that people are used to constant growth, but history includes accelerating growth; if that resumes through automation, centuries of change could be compressed into a short period. How the worldview could be wrong: He reviews key uncertainties: compute limits, software bottlenecks, whether AI can truly automate science, whether digital people are possible, and whether the right response is different from what he currently favors. Decision-making under weirdness: Holden argues that being hesitant because an idea sounds weird can be rational for many people, but not when your job is to search for high-upside, underexplored, world-shaping bets.
Key Arguments: Open Philanthropy needs separate leadership because global health/well-being and longtermist grantmaking now require fundamentally different organizational styles, scales, and bottlenecks. The longtermist case is not just a philosophical bet about future value; it is also an empirical claim that this century may contain decisive technological transitions, especially AI. If AI can automate scientific and technological advancement, then 100,000 years of progress could effectively become 10 years or less, creating an unprecedented period of instability and opportunity. Digital people are plausible because brains are physical information-processing systems; if we can reproduce the relevant computation, then software-based persons may be possible. If digital people are possible, they would likely be morally relevant persons, meaning the future could contain vast numbers of beings whose welfare matters greatly. Exponential or accelerating growth makes the current era look unusually weird from a broad historical or galactic perspective, even if day-to-day life feels normal. The strongest objections are not that the scenario is impossible, but that key estimates—compute, algorithms, training environments, timelines—could be wrong, and that the broader community lacks consensus. Holden thinks the right emotional stance is not hype or panic but stunned seriousness: the stakes may be enormous, but the correct responses are still unclear.
Data Points: Open Philanthropy grants recommended: over a billion dollars - Holden notes the foundation has recommended more than a billion dollars in grants. Longtermist funding scale: tens of millions of dollars - He says Open Philanthropy currently gives away tens of millions annually for longtermist causes. Projected century timescale: this century - The main thesis is that transformative AI and related shifts may arrive within the current century. Claimed acceleration scenario: 100,000 years could become 10 years - Used to illustrate how automating science and technological progress could compress long-term development. Estimated limit of current growth continuation: about 8,200 years - Holden says current economic growth rates cannot plausibly continue much longer at the same level without absurd physical implications. Economic growth under 2% annual rate: about 2% per year - He repeatedly uses 2% growth as a rough benchmark for modern long-run economic growth. Universe age: about 13–14 billion years - Used to place human history in cosmic perspective. Life on Earth age: a few billion years - Part of the timeline argument that our era is extremely early relative to possible future civilization. Space travel age: less than 100 years - Illustrates how recently humanity began technologies relevant to a galaxy-spanning future. AI field age: about 63–64 years - He argues the field is still relatively young, so forecasting remains highly uncertain. Time since first modern computers: 1945 onward - Cited to show how recent the key technological foundations are. Global health interventions example: $5 bed net - Holden references bed nets as a straightforward, evidence-backed charity intervention. Worldview investigation reports: multiple technical reports - He cites reports by Joe Carlsmith, Ajeya Cotra, Tom Davidson, and David Roodman as building blocks for the worldview.
Pivotal Quotes: "We just want to do the most good possible." — Holden Karnofsky (describing Dustin Moscovitz and Cari Tuna): Explaining the first meeting that led to Open Philanthropy’s longtermist funding relationship. "We live on this rocket ship that took off five seconds ago and nobody knows where it’s going." — Holden Karnofsky: His summary of why history looks extremely unusual when zoomed out across cosmic time. "The proper mood for the most important century is not being excited. It’s not being afraid. It’s just kind of being like, ugh." — Holden Karnofsky: He describes the emotional stance he thinks is appropriate toward the possibility of transformative change.
Implications: Listeners should expect huge uncertainty but also huge stakes: AI, digital persons, and future governance may be among the most consequential issues of the era. The takeaway is to think bigger, act earlier, and take longterm risk seriously without pretending the best solution is already known.