Position essay argues that accurate personal health records are crucial, prioritizing fidelity over mere storage.
As language models begin to out-diagnose physicians — one 2025 study reports roughly 80% accuracy on sequential diagnostic cases against about 20% for generalist doctors — this essay argues that medical knowledge has stopped being the binding constraint in personal health. The new bottleneck is older and humbler: a continuous, faithful record of a single life, the substrate any intelligence must reason over to be useful to one particular person. The essay makes a sharper, second claim: the hard problem is not storage but fidelity. A longitudinal record built from ordinary conversation must extract dated facts without distortion, because unfaithful memory compounds error and launders it into confident, misleading synthesis — worse than no memory at all. Fidelity, unlike raw capability, is measurable. The argument is stated as four falsifiable predictions (continuity over capability; fidelity over storage; demand preceding the product; the record, not the model, as the moat) and confronts the central objection: a record faithful enough to help is intimate enough to harm, so ownership must sit with the person. A position essay with references to Nori et al. (2025) and Costa-Gomes et al. (2026).
No takes yet. Share an insight, caveat, or question.
Rafael de Menezes Ehlers (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: