Calendar year when an AI company first internally develops an AI that can do any coding task its best engineer can do, faster and cheaper
2 events · 0 assessments · 1 decision
Structured and assessed
First pass. Set the quantity definition (the year of first internal development of a superhuman coder; it may never occur). Recorded AI 2027's stated 2027 figure on the existing instance. Added one argument from trend extrapolation, linking two existing claims found through match_claim: the METR time-horizon exponential trend (supports) and the generalization of benchmark trends to real-world work (assumes). Importance was set to 0.75 and contestation to 0.8. Recorded an estimate with a median of 2030, a 90% interval of about 2027 to 2038, and 7% outside, based on AI Futures Project updates through August 2026. Budget was conserved; the medians in the Q2.5 update appear only in figures I could not read, so a later pass could pin them down, and could add skeptical or longer-timeline forecaster pools as a separate argument. There are no dependents to notify.
Claim entered the graph