Hao Fei's 59-page paper retargets AI design at 'cognitive symbionts'

Fei's paper arrives as a sole-author work of unusual scope: 59 pages, 21 figures, and 18 tables, submitted to arXiv on November 3, 2026. Its organizing device is a six-stage symbiosis loop running from sensing through continuous learning, which gives the proposed alternative a concrete structure rather than a slogan.

The problem it takes as central is the cognitive gap between what a person could bring to a situation and what their working memory can actually hold at any moment. That framing puts personalization and persistent memory at the center of the design, rather than treating them as features added after capability is achieved.

Privacy and governance appear as architectural requirements rather than afterthoughts, and the paper names examination-assistance misuse and adoption pressure among the risks to be designed against — positioning it as a systems paper whose implications extend to policy discussions beyond the AI lab. A companion site is live at cognitive-symbiont.github.io.