Episode Summary
Executive Summary: Holden Karnofsky explains how GiveWell and Open Philanthropy evolved from trying to find the best charities to funding higher-risk, long-horizon opportunities like policy, scientific research, farm animal welfare, and AI safety. He emphasizes cause prioritization, hiring talented autonomous researchers, field-building, and using philanthropy to back neglected problems where small amounts of money can have outsized impact. The episode also covers OpenPhil’s hiring push and Holden’s thinking on utopian futures.
Main Topics: Origin and mission of GiveWell and Open Philanthropy: Holden traces the progression from GiveWell’s evidence-based charity evaluation to Open Philanthropy’s broader, more speculative grantmaking aimed at major neglected problems and long-term impact. Cause selection: importance, neglectedness, tractability: OpenPhil chooses focus areas by looking for large-scale problems that are overlooked and realistically solvable, using systematic research and hiring to shape priorities. Major cause areas and worldview split: The discussion covers OpenPhil’s work in criminal justice reform, farm animal welfare, global catastrophic risk reduction, scientific research, and EA support, plus the challenge of allocating funds across very different moral worldviews. Grantmaking process and organizational culture: Holden describes a decentralized grantmaking model centered on program officers, internal write-ups, probabilistic judgment, the 50-40-10 rule, and a culture of truth-seeking and disagreement. AI safety, timelines, and field building: OpenPhil sees advanced AI as a major long-term risk and opportunity, funding technical research, policy analysis, and career pipelines to build a durable AI safety field. Hiring and career opportunities at OpenPhil: A substantial portion of the interview focuses on roles being hired: generalist research analysts, AI specialists, operations, general counsel, and how applicants can grow into high-impact positions. Utopias and public attitudes toward ideal futures: Holden discusses an academic conference and an informal survey of utopian descriptions, finding that people prefer abstract freedoms and basic goods but resist highly specific visions of the future.
Key Arguments: Philanthropy is one of the few arenas where a small amount of money can directly change the world, often with far more leverage than individuals expect. GiveWell and Open Philanthropy differ in philosophy: GiveWell emphasizes proven, evidence-backed interventions, while OpenPhil pursues risky, neglected, long-horizon bets. The best causes often look unusual or counterintuitive because socially obvious causes already attract money and attention. History shows philanthropy can be hugely consequential, as with the Green Revolution and the development of the pill. Field-building matters because some important topics lack an institutional home; philanthropy can create the expertise pipeline needed for future breakthroughs. OpenPhil uses a decentralized model so the person closest to a cause has autonomy, but it balances this with oversight and a 50-40-10 portfolio standard. AI safety is a philanthropic priority because transformative AI could arrive relatively soon and create both accident risks and geopolitical/misuse risks. The biggest challenge in philanthropy is that evaluating worldviews and moral weights is not purely technical; it requires deep judgment about humans, animals, and future generations. Academia often rewards frontier advancement rather than synthesis, evaluation, or pragmatic decision support, leaving important neglected work for philanthropists to fund. Hiring well is central to impact because staff judgment and relationships determine both what grants get made and how effectively the organization learns.
Data Points: GiveWell-recommended funding flow: around $100 million a year - Money currently going to charities recommended by GiveWell Open Philanthropy annual giving: between $100 million and $200 million a year - Current scale of OpenPhil’s grantmaking Potential AI transformation timeline: at least 10% probability within 20 years - Holden’s estimate of transformative AI arriving soon enough to matter for current strategy OpenPhil staff size: about 20 people - Organization size at time of interview Planned research analyst hiring: 3 to 6 long-term research analysts - Holden’s medium-term expectation for the generalist hiring push Portfolio quality rule: 50-40-10 - Target mix for grant portfolios: 50% good, 40% okay, 10% discretionary Egregious incarceration study result: cutting incarceration could have no expected impact on crime - David Rudman’s evidence review cited in criminal justice reform work Utopian survey political skew: very heavily skewed left; Hillary Clinton would have won by 50 or 60 points - Description of the survey population for Holden’s utopia project Value of a gene drives grant: about $20 million - Funding aimed at speeding malaria-eradication technology Likely mortality benefit from malaria progress: around 1,000 lives per day - Holden’s estimate of deaths averted per day if eradication is sped up Farm animal welfare operational detail: a few cents a dozen eggs - Approximate cost to move from caged to cage-free egg production
Pivotal Quotes: "What I would urge someone to do if they were a billionaire and they were just getting started... would be to work really hard at those two decisions and take them slowly." — Holden Karnofsky: Advice on choosing causes and hiring before scaling philanthropy "I think one of the biggest mistakes to be made in hiring... is being someone who people won't tell the truth to." — Holden Karnofsky: Warning about how power dynamics can distort feedback to philanthropists "The world will be better off if there's already a large, robust, excellent field of experts who have spent their careers thinking very deeply about what could go wrong with AI." — Holden Karnofsky: Why OpenPhil funds AI field-building rather than only direct technical solutions
Implications: For donors and job seekers, the episode argues that impact comes from careful cause choice, talent, and long-term field building—not just passion. For philanthropy, it pushes a more analytical, contrarian, and portfolio-based model.