Toby Ord is an Oxford philosopher and existential-risk researcher. He is the author of The Precipice: Existential Risk and the Future of Humanity (2020) and of Broad Timelines (2026), and his work connects existential-risk philosophy with discourse on AI timelines.
Background
Ord is a Senior Research Fellow at Oxford and was a Fellow of the Future of Humanity Institute until its closure in 2024. The Precipice: Existential Risk and the Future of Humanity (2020) is his book on existential risk.
AI timelines positioning
Ord argues for wide probability distributions over AI timelines rather than committing to either a fast-takeoff or a slow-takeoff position. His core statement of this view is "Broad Timelines" (March 19, 2026), which flags the asymmetric costs of being wrong in either direction and advocates policy robustness across scenarios. Ord is a representative voice in the "epistemic humility" pole of AI-timelines discourse.
This wide-distribution framing serves as a counter-citation both to near-term-AGI positions such as those in Situational Awareness: The Decade Ahead, AI 2027, and Machines of Loving Grace, and to slow-takeoff positions such as Why I Think AI Take-Off Is Relatively Slow (Cowen) and AI as Normal Technology. Through both The Precipice and his timelines work, Ord links existential-risk philosophy with empirical AI-timelines discussion (AGI Timelines).
Notable analyses
Ord documented that o3-high attempts on ARC-AGI tasks (ARC Prize) cost roughly $30,000 per task, with about 43 million words generated, which he likened to "writing an Encyclopedia Britannica per task." The figure has been cited as an illustration of frontier reasoning-model inference economics.
Relationships
- authored: Broad Timelines
- related: Situational Awareness: The Decade Ahead (oppositional framing), AI 2027, Machines of Loving Grace, Why I Think AI Take-Off Is Relatively Slow (Cowen), AI as Normal Technology, AGI Timelines.