These questions form…: When will a frontier…

Part of: These questions form one register on recursive self-improvement (RSI) at frontier AI labs: whether, when and how fast AI systems come to automate AI research, whether anyone will be able to measure it, and how the handoff and pacing would be governed. Each row's background is self-contained and carries the verified baseline facts, definitions and instrument caveats; read it before researching. Attribution standard: where a line says someone thinks, claims, projects, estimates or models, that is evidence about the speaker, not about the world; several sources in this subject are themselves forecasts (AI 2027, AI 2040: Plan A, the AI Futures Model, Davidson-Houlden, Greenblatt's stated medians) and none may be adopted as a premise. Definitions: 'top-five AI developer by revenue' = the five largest frontier developers by AI revenue at resolution time (as of August 2026: OpenAI, Anthropic, Google DeepMind, xAI/SpaceXAI, Meta); 'frontier developer' = any developer of a model in the top ten on the Epoch Capabilities Index or the Artificial Analysis Intelligence Index; 'uplift' always means uplift IN VALUE (METR's convention), never coding throughput. Key verified baselines (August 2026): METR's Frontier Risk Report (2026-05-19) found companies 'did not report evidence of dramatic speed-ups' and that Anthropic 'had not seen a 2X increase in the pace of progress as of April 2026'; Anthropic reports >80% of merged code authored by Claude but 'well short of' a doubling of overall R&D progress; all three frontier safety frameworks contain AI-R&D-automation thresholds and all three labs declare them uncrossed; no jurisdiction regulates models as research instruments; METR's time-horizon suite is unreliable above 16 hours and its dashboard has been static since 2026-05-08. Reason through mechanism and current evidence; keep answers consistent across rows where they share a world; express genuine uncertainty as wide percentile ranges rather than false precision. — view all 16 rows

When will a frontier developer state publicly that a model it has released was designed principally by a preceding model of its own, rather than by its human researchers?✓#10

first_occurrence_dateJun 1, 2031

10% by
Jun '28
25% by
Sep '29
median
Jun '31
75% by
Jan '34
90% by
Dec '40

rationale:

The final estimate is anchored at P10: June 2028, P25: September 2029, P50: June 2031, P75: January 2034, and P90: December 2040.

Initial logic and parameters are validated as established context, including the technical capability distinction from implementation throughput 2 sources, the capability bottleneck in research taste 3 sources, and institutional disincentives surrounding automated R&D safety thresholds 2 sources.

Standard processing applied.

Pushing this timeline back ensures it strictly follows the lesser milestone of AI selecting a majority of experiments . Jumping directly to the final transformation, the timeline lands on a June 2031 median, with institutional caution and safety frameworks stretching the upper percentiles out to December 2040.

Set against related questions, this timeline was pushed back to ensure it strictly follows the lesser milestone of AI selecting a majority of experiments .

Ask a followup

Ask a followup