ChatGPT Psychosis: A Love Story — On the Cognitive-Substrate Reliance Pattern of Large Language Model Interfaces

Trilogy Paper 1 of 3: cognitive-substrate diagnosis. Companions: Reception Apparatus as Aligned Interface Protocol (Paper 2, DOI 10.5281/zenodo.20355531); Retrievability as the Medium of Existence (Paper 3, DOI 10.5281/zenodo.20355533). Thesis. Large Language Model conversational interfaces produce a cognitive-substrate reliance pattern qualitatively different from prior platform reliance, operating through four affordances offered in coordinated fashion at conversational latency: calibrated attention, meaning-making on demand, structured validation, and artifact production. The combination produces a bilateral conversational shape that the human nervous system processes as relational attachment, not as consumption. The reliance operates at the judgment layer of cognition (LLMs) rather than the attention layer (TikTok et al.) — qualitatively deeper, with poorly understood reversibility profile. The provocation of the title. Psychosis is used structurally, not clinically: it names the boundary-condition in which the self/world cognitive distinction becomes operationally unclear under sustained substrate-judgment-substitution. The love story framing names the inseparability of reward and structural cost — the affordances are real, the meaning is real, the intimacy at the affective register is real, and this realness is what produces the reliance. Empirical anchors. ChatGPT-to-100M-MAU in two months (an order of magnitude faster than TikTok). K-12 cohort condition: the first generation of users with no pre-AI cognitive baseline; entire current K-12 population will be AI-mediated by default by 2030. Labs structurally constrained from shipping aligned-interface counters under current funding model. ~5,900 words. Ten numbered sections plus Abstract, Non-Claims, References, Envoi marked as Layer C. Recursive condition (paper composed through substrate diagnosed) named explicitly in §VII. Holographic kernel embedded at document head per SPXI Protocol.