Philanthropists tend to choose causes based on personal experience, geographic proximity, or emotional appeal. This approach is understandable — personal connection is what motivates many people to give in the first place. But some causes offer far greater opportunities for impact than others: the same donation might save ten lives in one place but a thousand in another.
One of our core values is our tolerance for philanthropic risk. We’ve seen that the biggest philanthropic wins often come from ideas that seem unlikely to succeed, so we’re open to funding a lot of work that could fail in order to find a few transformative successes.
When choosing which causes to support, we face difficult moral and empirical questions without clear "right" answers. To address this uncertainty, we divide resources across several broad approaches to doing good, each grounded in a different worldview.
This program funds projects that use AI to improve probabilistic forecasting and support sound decision-making. We aim to support the creation of AI models that enhance forecasting accuracy, tools that help with tasks relevant to clear reasoning, and research to understand when models do and do not support truth-oriented reasoning.
Editor’s note: This article was published under our former name, Open Philanthropy. Open Philanthropy program staff often create back-of-the-envelope calculations (BOTECs) as they decide whether to make a grant. This is especially common for areas in our Global Health and Wellbeing (GHW) portfolio. We introduced BOTECs in a recent post:...
If you ask Abel Brodeur about his interest in p-hacking — manipulating data or analysis to get a statistically significant result — he'll be honest with you.
Our Potential Risks from Advanced AI program is now called Navigating Transformative AI. While the vast majority of work we fund is still aimed at catastrophic risk mitigation, the new name better captures the full breadth of what we aim to support: work that helps humanity to successfully navigate the transition to transformative AI.
Many experts predict that leading AI systems will become smarter-than-human in the next decade, but efforts to mitigate the risks remain profoundly underfunded.
Our AI safety strategy has evolved significantly since 2015, moving from early field-building to a comprehensive three-pillar approach: improving visibility into AI capabilities through evaluations and forecasting, developing technical and policy safeguards against catastrophic risks, and building the talent pipeline and institutional capacity the field urgently needs.
