These questions form…: As of 2026-08-13,…

Part of: These questions form one register on recursive self-improvement (RSI) at frontier AI labs. Each row asks for a percentage share (0-100); read each row's self-contained background before researching - it carries the verified baseline facts, definitions and instrument caveats. 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 or model outputs and none may be adopted as a premise. Definitions: 'largest AI lab by revenue' and 'top-five AI developer by revenue' refer to the largest frontier developers by AI revenue at the relevant date (as of August 2026: OpenAI, Anthropic, Google DeepMind, xAI/SpaceXAI, Meta). Most rows are NOW-CASTS of latent quantities that no lab publishes and no tracker carries (share of experiments proposed by AI, share of RL environments authored by AI, share of compute on experiments, share of safety effort on automated-R&D risks): estimate what is TRUE, not what has been documented, and express genuine uncertainty as wide percentile ranges. Key verified baselines (August 2026): METR's Frontier Risk Report (2026-05-19) found companies 'did not report evidence of dramatic speed-ups', that Anthropic 'had not seen a 2X increase in the pace of progress as of April 2026', and that METR is 'not aware of evidence that any company relies on AI agents for setting research agendas'; Anthropic reports >80% of merged code authored by Claude; OpenAI's GPT-5.6 launch post said the share of research compute devoted to internal coding inference grew 100-fold in six months; METR reports at least 16% of successful long-task runs were illegitimate on review in Time Horizon 1.1. Keep rows consistent with each other. — view all 5 rows

As of 2026-08-13, what share of frontier-lab alignment and safety research effort, measured in full-time-equivalent researcher-years, addresses risks arising specifically from AI systems conducting AI research, rather than risks from deployed models?✓#5

share14.5percent (0-100)

rationale:

The central estimate places this share around 14.5%, with a p10 of 6.0%, p25 of 10.0%, p75 of 22.0%, and p90 of 32.0%. Initial logic and parameters are validated. The estimate is consistent with the broader view that internal AI-R&D risk remains a minority focus compared to external deployment risks 3 sources. Standard processing applied to individual lab portfolios. Jumping to the final transformation: the internal-agent risk portfolio is undeniably growing 66 sources. The wide interval - spanning from roughly 6.0% at the low end to 32.0% at the high end - reflects two primary uncertainties. First is the definitional boundary of dual-use safety work anthropic.com. Second, unpublished internal monitoring and control programs at OpenAI and Anthropic may be larger than public evidence suggests, supporting the right tail.

This estimate held up because it was already consistent with the broader view that internal AI-R&D risk remains a minority focus compared to external deployment risks.

Ask a followup

Ask a followup