Google recently decided to remove inaccurate AI-generated health summaries from its search results. While OpenAI launched ChatGPT Health gving users access to health and wellness information, its service terms state that it is not intended for medical diagnosis or treatment. Despite AI companies venturing into the health and wellness sector, it is limited when it comes to professional medical expertise. Something users seem to wilfully ignore.
Dr Jess Morris, general practitioner (GP) at Mediclinic Morningside, says many people are increasingly turning to AI to help interpret symptoms or test results before consulting a doctor.
“In some cases, this can cause unnecessary stress, but in others it can create false reassurance, delaying professional care when early intervention could make a critical difference.”
Morris unpacks three common health concerns that even the most sophisticated AI cannot reliably address.
Blood pressure
The challenge with high blood pressure, or hypertension, is that it often lacks any obvious symptoms.
As Morris points out, many people only discover they have dangerously high readings during routine check-ups for other unrelated health matters.
“By the time symptoms such as chest pain, blurred vision or anxiety appear, a person may already be experiencing a hypertensive crisis,” he explains.
“A single reading, or an online reference range, cannot show how blood pressure behaves over time or during sleep, nor can it reflect an individual’s risk of stroke or heart disease.”
Cholesterol
Cholesterol levels present a similar problem.
Online resources and AI chatbots can explain what cholesterol is and outline general thresholds, but they cannot assess the balance between different types of cholesterol or determine how those levels interact with personal risk factors.
“Many people with abnormal lipid levels feel perfectly well until they experience a heart attack or stroke,” notes Morris, who adds that doctors consider cholesterol results alongside family history, weight, age and other key conditions such as diabetes to decide whether lifestyle changes are sufficient or whether medication is needed.
Prediabetes
Prediabetes is another area where oversimplified answers can often fall short.
“Blood glucose levels that sit slightly above normal do not automatically mean someone will develop diabetes, nor do they guarantee safety,” he explains.
“Prediabetes often has no symptoms and is typically identified through screening tests such as HbA1C, which reflects blood sugar levels over several months rather than at a single point in time.”
Clinicians interpret these results together with weight changes, activity levels and previous test outcomes to build an accurate picture.
Importantly, prediabetes is reversible, but only when it is correctly identified and managed.
Not medi-telligent
These three examples highlight why there is rarely a universal norm in healthcare.
“Medical test results are pieces of a larger puzzle, not definitive answers in isolation,” he says.
“While AI tools can help people access general information, they are not equipped to replace clinical judgement or personalised care.
"When it comes to understanding your health, there is no shortcut that replaces a conversation with a qualified healthcare professional who can consider the full context and guide you towards appropriate care,” Morris concludes.
