LessWrong (Curated & Popular) "Broad Timelines" by Toby_Ord
11 snips
Mar 21, 2026 A clear look at deep uncertainty about when AI will transform the world. Definitions and contrasting short versus long timeline views are discussed. The conversation surveys expert probability distributions and shows why single-year forecasts mislead. Practical planning under uncertainty, hedging toward early risks, and building mixed portfolios of short- and long-term work are explored.
AI Snips
Chapters
Transcript
Episode notes
Adopt Broad Distributions Over AI Timelines
- Expert disagreement implies broad uncertainty over AI timelines rather than a single short or long date.
- Tobias H argues the correct epistemic response is a wide probability distribution because many informed peers and disciplines disagree.
Experts Show Highly Skewed Wide Forecasts
- Individual expert forecasts are often highly skewed with wide 80% intervals spanning decades.
- Tobias H highlights Daniel Cocotaljo's distribution peaking in 2028 but with a median of 2030 and an 80% window from 2027 to beyond 2050.
Single-Year Predictions Invite Misinterpretation
- Communicating a single year for arrival risks public misinterpretation and reputational harm.
- Tobias H warns the 2027 scenario was widely reported as a sharp prediction, which can lead to false discredit when that year passes.
