this past week, openai launched chatgpt health.
chatgpt health will live inside the standard chatgpt interface as a special “health” tab or mode. in practice, it works like a normal chatgpt chat, with two main differences:
you can link your medical records, sync apple health stats, or connect wellness apps like myfitnesspal (diet tracking) and weight watchers.
it runs in a walled-off, privacy-enhanced environment within chatgpt
surprisingly (or not), my x feed was harshly negative.
the top concerns:
safety and “hallucination” risk: llms don’t guarantee correct or safe medical advice. chatgpt generates the statistically likely answer, not necessarily the clinically correct one
privacy and data security: there’s nothing structurally preventing openai (or partners) from mining health queries for ads or other purposes later on. breaches or leaks could expose extremely personal details
re (1): i think it’s valid to worry about safety and hallucinations. but people will need to learn to treat llm advice like everything else on the internet: with a grain of salt. if you follow reddit health advice and get sick, people think you’re an idiot. same standard should apply here. for well-adjusted adults, ai healthcare should be just another tool in the wellness toolkit.
re (2): if, however, you’re a highly paid tech employee who’s anti-chatgpt health primarily for privacy reasons, i think you’re wrong, and you should feel bad about it.
globally, over half the world (4.6 billion people) lacks access to essential health services (who)
over 100 million americans (nearly one-third of the country) don’t have regular access to primary care due to provider shortages (nachc)
even where clinics exist, one-third of americans said they couldn’t see a doctor when they needed to, due to availability problems (who)
in 2023, 26 million americans had no health insurance at all (commonwealth)
even with insurance, 36% of american adults skipped or postponed care last year due to cost (kff)
these are devastating statistics. healthcare is an essential human need that is catastrophically undersupplied. billions of people globally need better care. ai will help distribute higher quality health information globally, for basically no marginal cost. so if you’re some tech dork or think girl who’s bashing chatgpt health because it’s “creepy,” i think you’re completely out of touch.
in my book, ai for healthcare is a clear-cut win for mankind.