These two rows replace…: When will OpenAI,…

Part of: These two rows replace anonymised versions in FutureSearch's published register on recursive self-improvement, naming the three developers that actually publish frontier-safety frameworks with AI R&D thresholds. Where a source says someone thinks, claims, projects or estimates, that is evidence about the speaker, not about the world. If the event is very unlikely to occur before 2047, express that in the upper percentiles. — view all 2 rows

When will OpenAI, Anthropic or Google DeepMind state publicly that its AI research organization no longer requires human researchers for a substantial category of its work?✓#2

first_occurrence_dateSep 1, 2033

10% by
Sep '28
25% by
May '30
median
Sep '33
75% by
Jan '39
90% by
never

rationale:

Final Estimates: 10th percentile at September 2028, 25th percentile at May 2030, median at September 2033, 75th percentile at January 2039, and 90th percentile at January 2049. The Resolution Standard and Current Status Initial logic and parameters are validated. Capability Ambitions vs. Reality Standard processing applied. Structural Disincentives for a Public Declaration Standard processing applied. Trajectory and the Long Tail Consistent with an updated assessment of the likelihood that labs will officially claim full automation of key research categories before 2032 , the trajectory shifts directly to the final outcome. Median resolution lands at September 2033, with the 75th percentile at January 2039 and the 90th percentile at January 2049.

Shifted the distribution slightly later to remain consistent with an updated assessment of the likelihood that labs will officially claim full automation of key research categories before 2032 .

Ask a followup

Ask a followup