Artificial intelligence is already reshaping healthcare behind the scenes — from clinical decision support to drug discovery and hospital administration. But AI’s influence is now increasingly extending beyond professionals and institutions, and directly into the hands of patients navigating increasingly complex health systems.
A new report from OpenAI offers a data-heavy look at how people are using ChatGPT for health-related questions, adding a patient-driven layer to a trend investors and healthcare executives have been watching for several years. The findings highlight how AI’s impact in medicine is increasingly visible in access, administration and self-advocacy.
From clinical tool to patient companion
According to the report, more than 5% of all ChatGPT messages globally now relate to healthcare, with more than 40 million health-related prompts submitted every day. In the US alone, roughly one in four regular users ask health questions each week.
Importantly, most of this usage is not diagnostic in nature. Instead, people are turning to AI to interpret medical information, prepare for appointments, understand insurance coverage and manage the administrative sprawl that often follows clinical care.
Health insurance is a dominant theme, driven largely by a US healthcare system that the report calls a “long-standing and worsening pain point for many”. OpenAI estimates between 1.6 and 1.9 million ChatGPT messages each week focus on comparing plans, understanding pricing, dealing with claims and appeals, and navigating eligibility and billing. For industry observers, that points to a familiar bottleneck — administrative complexity — but highlights a new workaround emerging at the consumer level.
Filling gaps when systems fall short
One of the report’s more striking findings is when people are seeking help. Around 70% of health-related conversations occur outside normal clinic hours, reinforcing the idea that AI is being used as a first-line information source when traditional care is unavailable.
Geography is another important factor. Users in underserved rural communities generate nearly 600,000 healthcare-related messages each week, with particularly high usage in so-called “hospital deserts” — areas more than 30 minutes from a general medical facility.
The report is explicit about AI’s limits: it cannot reopen closed hospitals or replace specialist services that have vanished from regional areas. But it can help patients interpret symptoms, organise information and decide when escalation is necessary — functions that sit upstream of formal care.
For healthcare systems already under strain, this kind of informal triage and preparation could have meaningful second-order effects, potentially improving how — and when — patients engage with clinicians.
Self-advocacy as a use case
Several case studies in the report illustrate a recurring theme: patients using AI to advocate for themselves within rigid systems.
In one example, a US patient with a rare autoimmune condition used ChatGPT to compile and summarise peer-reviewed literature supporting an insurance appeal for a denied therapy — an appeal that ultimately succeeded. In another, the tool helped flag potential medication interactions and organise lab trends between appointments.
These anecdotes are not evidence of clinical efficacy, but they do reflect a shift in how information asymmetry is being addressed. AI is lowering the barrier for patients to understand — and question — decisions that were once opaque.
For biotech and health-tech investors, that has implications for everything from patient engagement strategies to reimbursement dynamics.
Professionals are already on board
The patient story runs parallel to — rather than instead of — professional adoption. The report notes that around two-thirds of US physicians used AI for at least one task in 2024, up sharply from the year before. Nurses, pharmacists and administrators are also regular users, primarily for documentation, workflow support and information retrieval.
Crucially, diagnostic use remains limited and relatively stable, suggesting AI’s near-term value lies more in efficiency and support than in replacing clinical judgement.
What this means for healthcare investing
Taken together, the report reinforces a broader message: AI in healthcare is not a single market or product category. It spans infrastructure, software, services and now consumer-level tools that sit alongside, rather than inside, formal care pathways.
The message isn’t that AI will “solve” healthcare’s problems, but that adoption is becoming more distributed, with patients increasingly participating in how data, information and decisions flow through the system.