Found first: a primary source the press has not covered yet.
Hao Fei's 59-page paper "From Autonomous Agents to Cognitive Symbionts: Toward AI-Augmented Human Intelligence", submitted to arXiv on November 3, 2026, argues that the dominant push toward artificial general intelligence is aimed at the wrong target. The paper names an alternative: AI-augmented human intelligence (AHI), built on what Fei calls "cognitive symbionts," systems designed to be lifelong, embedded, real-time, private, low-friction, and proactive companions for individual people.
What the source says
The paper is a sole-author work spanning 59 pages, 21 figures, and 18 tables. It organizes the design space as a six-stage symbiosis loop: sensing, contextual retention, understanding, task execution, collaborative thinking, and continuous learning. The central problem it addresses is the cognitive gap between what a person could bring to a situation and what their working memory can actually hold at any moment. Privacy and governance are framed as architectural requirements, with the paper explicitly naming examination-assistance misuse and adoption pressure among the risks to be designed against. A companion site is live at cognitive-symbiont.github.io.
Why it matters
The paper's contribution is definitional. It names a design target, AHI, with a concrete six-stage loop and specified architectural properties, shifting the frame from autonomous capability to individual augmentation. That shift has practical consequences: it foregrounds personalization, persistent memory, and embedded context as first-order requirements. The governance section's treatment of misuse and adoption pressure as design constraints positions this as a systems paper with policy implications beyond the AI lab.