Harsh Jangid is Director at Coozmoo, an AI-powered digital marketing agency that helps SMBs accelerate growth and increase revenue.gettyFor most industries, zero-click search is a marketing headache. A user asks an AI assistant a question, gets a synthesized answer and never clicks through to a website. Traffic dips, attribution gets murky, marketing teams scramble to adapt. Annoying, but survivable.In healthcare, zero-click search leads to questions about who gets treated correctly, who gets treated at all, and who gets hurt by advice a machine delivered with total confidence and partial accuracy.That distinction, between an inconvenience and a hazard, is the one the AI industry has been slow to draw, and it's the one healthcare leaders need to draw urgently.The Mechanic Is The Same Everywhere: The Stakes Are NotZero-click search has spread fast and is no longer confined to Google's AI Overviews. It runs across ChatGPT, Perplexity, Gemini, Copilot and Meta AI, with users increasingly asking a question and getting a single answer instead of a page of links to evaluate themselves. That shift, from retrieval to synthesis, is reshaping how every industry gets discovered.For a retailer, that shift means fewer clicks to the product page. For a restaurant, it means an AI assistant naming a competitor instead of you. Frustrating, but the consumer's downside if the AI gets it wrong is a bad dinner or a missed sale.For healthcare, the downside is different in kind, not just degree. When someone asks an AI assistant, "I've had shortness of breath for two days, should I be worried?" the model isn't just competing for a click. It's making a triage decision, whether it means to or not.What The Evidence Actually ShowsA 2026 audit published in BMJ Open tested five major chatbots—ChatGPT, Gemini, Meta AI, Grok and DeepSeek—against 50 common health questions each. Roughly half of all responses were judged problematic, with close to one in five rated highly problematic.Separately, the largest user study of LLMs assisting the public with real medical decisions, led by Oxford's Internet Institute and Nuffield Department of Primary Care Health Sciences, published in Nature Medicine, found a troubling gap: These models now perform well on standardized medical knowledge tests, yet still pose real risks to everyday users trying to get help with their own symptoms. None of this means AI is useless for health questions. It's often genuinely good at translating a discharge summary. The risk isn't AI touching healthcare. It's AI being trusted for the one job, acute triage and diagnostic judgment, it's least reliably good at, precisely because it delivers every answer, right or wrong, with the same fluent confidence.Why ‘YMYL' Exists—And Why It's About To Matter More, Not LessLong before generative AI, search engines had a name for this category of risk: YMYL, or "Your Money or Your Life." It's the classification search quality systems use for content where being wrong carries real consequences, such as medical guidance, financial advice, legal information and safety instructions. YMYL content has always been held to a stricter bar than a recipe blog or a product review, because the cost of a confident wrong answer is categorically higher.Zero-click AI search inherits that risk category but strips out a safeguard that came with it. The old model preserved a ranked list a person had to read and weigh. That friction was also a filter. A synthesized AI answer removes it, the user gets one voice, sounding exactly as certain when it's wrong as when it's right.This is precisely why the same logic search engines apply to YMYL content, demanding stronger signals of expertise, authoritativeness and trustworthiness before surfacing something, has to carry over into how AI platforms decide what to say, and how confidently, when a person describes a symptom rather than a search term. RankRabbit AI analyzed and found that across more than 350,000 business profiles, only a small fraction were ever surfaced by AI assistants when users asked for recommendations. AI systems, it turned out, were evaluating confidence, not just relevance, and excluding anything they couldn't verify. That instinct, better to say nothing than to recommend something unverified, is exactly the instinct AI systems need to apply more consistently to clinical guidance itself, not just to which provider gets named.Trust Is Now Assigned Before Any Human ContactThere's a second-order effect that gets less attention than the accuracy problem. Increasingly, the AI isn't just answering the medical question, it's also naming who to see about it. A patient describes back pain radiating down a leg, and the assistant doesn't hand back a list to evaluate. It names a specialty, sometimes a specific clinic.That means trust in a provider is being assigned earlier in the process than it used to be, before a phone call, before a website visit, before any human judgment enters the picture at all. A clinic with a thin digital footprint, inconsistent listing data or few recent reviews can be functionally invisible to these systems, regardless of the quality of care it actually provides. The short version: The systems aren't ranking pages anymore, they're evaluating whether a provider is legible enough, across search content, reviews, social presence and directory data, for a cautious system to vouch for.The Bottom LineEvery industry watching zero-click search reshape discovery is having some version of the same conversation: How do we stay visible when the answer engine gives one response instead of 10 links? Healthcare is having that conversation, too, but it's not the most important one happening in the room. The more urgent question isn't whether a clinic shows up in an AI's answer. It's whether the answer itself, delivered with total fluency to someone who is scared and doesn't know what's wrong, is actually safe.Search used to hand people a shelf of books and let them decide which to trust. AI hands them one book, already open to a page, and says "read this." In most categories, that's a convenience. In healthcare, it's a responsibility the industry hasn't fully accepted yet, and the data so far suggests it should.Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Zero-Click Search Was An SEO Problem; In Healthcare, It's A Patient-Safety Problem
The distinction between inconvenience and hazard is the one the AI industry has been slow to draw, and it's the one healthcare leaders need to draw most urgently.
1,015 words~5 min read
