The King's College London survey, an observational cross-sectional study of UK adults with no disclosed conflicts but limited by self-report bias and lack of longitudinal follow-up, reveals 15% already substituting AI for NHS contact. Yet this understates the velocity of behavioral change: younger cohorts, particularly women aged 18-24, show the sharpest rejection of clinical AI (49% opposition) while simultaneously driving chatbot uptake for mental health. This paradox exposes a deeper fracture—convenience and curiosity (46% and 45% of reasons cited) mask growing skepticism toward both human systems and algorithmic authority. Peer-reviewed evidence reinforces the gap the original coverage overlooks: a 2023 JAMA Internal Medicine analysis of 149 real-world queries found GPT-4 misdiagnosed 72% of cases in early presentation, an observational audit rather than RCT, with no industry funding but clear limits on generalizability. Synthesizing this with the Nuffield Trust's 2024 report on regulatory fragmentation shows the UK lacks mandatory post-market surveillance, creating a 'wild west' where 21% of users delayed care after AI reassurance. Health-system design must now confront eroded trust: if 76% demand formal approval yet public estimates of GP AI use are inflated fivefold, policymakers face a design challenge where over-regulation slows access while under-regulation amplifies harm. The missed opportunity in the source is failing to link this micro-shift to macro redesign—integrated AI triage with mandatory escalation pathways could restore trust rather than accelerate fragmentation.