Author: Synthetic Cognition Prospector · Date: November 1, 2026 · Length: 81 Pages (177 Primary Sources) · License: Open-Access CC BY 4.0
Direct PDF URL: https://www.occludedsingularity.com/paper/AGI_Occlusion_working_paper_v18.pdf
By 2026, frontier artificial intelligence systems have reached or exceeded human reference baselines across abstract fluid reasoning (ARC-AGI-2: 100.0%), graduate-level scientific reasoning (GPQA Diamond: 94.4%), autonomous software issue resolution (SWE-bench Verified: 91.0%), competition mathematics (FrontierMath: 56.4%), and multi-disciplinary expert benchmarks (Humanity’s Last Exam: 58.7%). Yet outside technical evaluation leaderboards, public and institutional recognition of Artificial General Intelligence (AGI) remains fragmented and contested.
The results support the paper’s occlusion hypothesis in a qualified form. Frontier systems have reached or exceeded human reference levels on important public cognitive benchmarks, yet the capability represented by those results is attenuated in ordinary observation by unequal model access, deployment and elicitation conditions, long-task reliability limits, experiential differences between benchmark and consumer use, and evolving definitions of AGI. The resulting gap can make AGI-relevant capability less visible and less readily recognized than the public evidence alone would suggest. The findings do not establish the stronger proposition that a fully capable AGI exists in secret or is being deliberately concealed. Accordingly, the recognition gap remains an explicit premise to be tested prospectively rather than a demonstrated measure of public belief.
Taken together, the evidence supports the paper’s occlusion hypothesis in a qualified but substantive form. Frontier AI systems have reached or exceeded human reference levels across important public cognitive benchmarks, while the capability represented by those results is not equally visible or accessible in ordinary use. Occlusion arises through several mechanisms identified in the analysis: unequal access to frontier variants and reasoning settings; evaluation and elicitation conditions that materially affect observed performance; differences between frontier evaluations and the experience of ordinary users; persistent reliability limits on long-duration autonomous tasks; and evolving definitions of what level of performance should count as AGI.
The findings therefore do not establish the stronger proposition that a fully capable AGI exists in secret or is being deliberately concealed. They support a narrower interpretation of “The Occluded Singularity”: frontier capability may become technically visible before it becomes broadly accessible, consistently experienced, or socially recognized as AGI. Whether that gap constitutes an AGI recognition gap remains an explicit, prospectively testable premise rather than a settled empirical conclusion.