Is it True That One in Seven People Use AI Instead of a GP?

The rise of AI health advice tools, notably generative AI chatbots powered by large language models (LLMs), has stirred curiosity and concern. Headlines suggest that "one in seven people" now rely on AI assistants instead of visiting their general practitioners (GPs). But what’s behind these claims? Is this shift real, and what does it mean for trust in AI assistants compared to traditional health consultations?
The Growing Popularity of AI Chatbots for Health Questions
For years, many turned to search engines — affectionately or critically dubbed "Dr Google" — for quick health information. However, traditional search results can overwhelm or mislead, as they often present a jumble of credible and dubious sources side-by-side.

The latest trend moves away from multi-link lists to conversational AI chatbots systematic review bias assessment https://www.cochrane.org/about-us/news/why-evidence-synthesis-matters-age-ai that generate instant, human-like responses. These generative AI tools promise to synthesize information and answer complex questions on the fly, seemingly streamlining health inquiries.

Some surveys and reports suggest that nearly 1 in 7 people have chosen to consult AI chatbots for health advice instead of booking an appointment with a GP. While this figure attracts attention, it's critical to examine the evidence carefully, and understand nuances behind usage patterns, motivations, and limitations.
From Search Engines to Generative AI Chatbots Search engines: Provide links to diverse sources, requiring users to interpret and verify information themselves. Generative AI chatbots: Use large language models (LLMs) trained on massive text datasets to generate direct answers in conversational form.
The shift implies users expect faster, clearer, and more conversational health information. But is this move from search to dialogue truly replacing GP consultations?
Large Language Models: Predicting Text, Not Truth
Understanding how LLMs work is key to appreciating their strengths and weaknesses. Contrary to some misconceptions, these models do not "understand" health or verify facts. Instead, they predict the likeliest next word sequences based on patterns in their training data.

This means that, while LLMs can produce coherent and sometimes impressively accurate health information, they also have a known tendency to generate hallucinations — confidently phrased but incorrect or entirely fabricated statements.

Of particular concern are occasions when AI generates invented citations or references, giving the illusion of evidence-based answers without real backing. This can mislead users who are not trained to spot such errors and may take AI outputs at face value.
Why Hallucinations Matter in AI Health Advice False reassurance: Incorrect or fabricated info might delay seeking needed care. Undetected harm: AI chatbots often don’t cover potential risks or uncertainties adequately. Damaged trust: Discovering AI mistakes may erode confidence in digital health tools broadly. Trust as the Limiting Factor in AI Health Assistants
Trust remains the central challenge for AI health advice to be safely integrated alongside traditional care. While generative AI chatbots offer convenience, users, clinicians, and regulators worry about reliability, accountability, and transparency.

Organizations dedicated to evaluating health evidence underscore the importance of trustworthy information. For example, The Cochrane Collaboration, an international network of researchers and healthcare professionals, systematically examines medical evidence to ensure decisions rest on solid foundations.

The Cochrane Library provides open access to high-quality systematic reviews — a gold standard for summarizing evidence. These reviews help patients and clinicians make informed choices, contrasting with AI outputs that may lack rigorous sourcing or context.
The Role of Trusted Sources in the Age of AI Reliable health advice depends on evidence synthesis like that produced by Cochrane, not just text generation. Human oversight remains vital to interpret complex information, assess risks, and contextualize symptoms. Trust in AI assistants hinges on transparency about their limitations and the nature of their outputs. Is One in Seven Really Using AI Instead of GPs?
Quantifying how many people prefer AI assistants over GPs is tricky. Surveys report varying numbers depending on region, age group, and definition of “use.” Some studies have claimed figures near or above 1 in 7, but these often reflect:
People who tried AI chatbots at least once, rather than consistent replacement of GP visits. Users supplementing rather than substituting professional care. Self-selected samples more engaged with digital health tools.
No credible evidence suggests a wholesale replacement of GPs by AI chatbots yet — certainly not a majority. What’s clearer is that many are experimenting with AI for quick questions, symptom checking, or second opinions, sometimes delaying care or seeking reassurance.
What Would Change Our Minds?
To fully accept the claim that "one in seven people use AI instead of a GP," we would want:
Rigorous, large-scale, and representative population studies measuring patterns of AI and GP usage over time. Data differentiating casual AI use from deliberate substitution of medical consultation. Peer-reviewed analysis that accounts for demographic and contextual factors influencing health decision-making.
Until then, such statistics should be treated cautiously and not overinterpreted as proof of widespread substitution.
How to Navigate AI Health Advice Safely
Given the realities above, here are practical tips for patients and clinicians:
Use AI assistants as starting points, not final diagnoses. Treat chatbot outputs as informational, not prescriptive. Verify AI suggestions using trusted sources like the Cochrane Library. Access systematic reviews for the best evidence summaries. Consult healthcare professionals for persistent, severe, or complex symptoms. AI can’t replace clinical judgment or physical examination. Be wary of confident-sounding AI statements that lack verifiable sources. Watch out for invented citations or overconfident advice. Advocate for transparency. AI developers should clearly communicate their tools’ capabilities and limitations. Conclusion
While AI chatbots based on large language models have revolutionized how some people seek health information, the claim that one in seven people use AI instead of a GP should be interpreted cautiously. These tools predict text but do not possess understanding or truth validation, leading to risks such as hallucinations and fabricated citations. Trust is paramount — and it must be built on transparent, reliable evidence, like that curated by The Cochrane Collaboration and the Cochrane Library.

AI health advice tools complement but do not yet replace the nuanced, evidence-based care provided by healthcare professionals. Thoughtful use, paired with critical appraisal and trusted research, remains the safest path forward.