When people first said that "doctors are just a tap away," the phrase typically referred to the growing ease of booking appointments or consulting specialists remotely. However, with artificial intelligence taking over everyday life, that idea has taken on an entirely new meaning.
What could be more convenient than describing a symptom and getting an instant diagnosis?
Bengaluru-based software engineer Manik Deepak is among those who rely on AI for health-related queries.
"Convenience. They have proper domain knowledge and it's very convenient to describe your symptoms. Rather than going to a doctor, getting an appointment, spending money, you can do all that at home for free," he says.
But not everyone shares that instinct. Suyog Shetti, despite being well acquainted with AI tools and chatbots, says it has never occurred to him to ask ChatGPT to diagnose his symptoms.
Between these two perspectives lies a growing behavioural shift that doctors say they are increasingly encountering in clinics.
As artificial intelligence becomes an increasingly common source of health information, questions are emerging about its role in the diagnostic journey.
What is driving the pattern of AI consultations? And what do doctors have to say about it?
The Appeal of Instant Answers
For many users, the biggest attraction is convenience. Shraddha Dayanand says she turns to AI chatbots for minor health concerns, mainly to understand how long symptoms may last before things return to normal.
"If it's something that I feel is very trivial or I know it's going to go away, that is when I use a chatbot," she says.
"It's easier," says Anandeshwar, who occasionally asks chatbots about health concerns. The last symptom he searched for was a mouth ulcer.
A broader pattern is beginning to emerge. People often turn to AI with symptoms they believe are manageable or to determine whether they are serious enough to warrant a doctor's visit.
When Neha Goswami feels unwell, her first instinct is not always to book a doctor's appointment.
"First I ask whether it is anything serious or dangerous," she says.
However, for users like Manik, who have great faith in AI's potential, the interaction continues throughout the recovery process.
"Before seeing a doctor, also, after coming back from the doctor. I will tell what doctor has suggested and then throughout the recovery phase, I will keep updating the chatbot to see how my progress is going."
On average, respondents rated their satisfaction with AI-assisted diagnosis at 6.75 out of 10.
A Doctor Who Never Judges
Consulting an AI chatbot often involves sharing highly personal information. Yet, many users appear comfortable doing exactly that.
The reason is not ignorance, but comfort.
Several respondents said they disclose extensive information, including age, medications, lifestyle factors and family medical histories.
Manik says he provides information "down to the last minute detail".
He is also unconcerned about privacy.
"I am not scared of my data being leaked or used for training models. I'm a software engineer. I don't mind it."
For others, the appeal lies elsewhere.
"I feel like the doctor might judge, but a robot won't," says Shraddha.
Many users also upload detailed laboratory reports. Interpreting medical reports can often leave patients confused, hoping someone can explain them in language they understand.
Neha Goswami says she uploaded her father's medical report to better understand his condition. Manik also regularly uploads laboratory reports for detailed explanations. And while Shraddha hasn't uploaded a report yet, she says she would feel comfortable sharing one.
"That's the best part about a chatbot," says Manik. "You can ask it to explain it in terms of quantum physics or maybe explain it to a five-year-old."
Anandeshwar echoes the same sentiment.
"I ask it to keep the language simple, and it follows that."
Reassurance Machine or Anxiety Machine?
On the surface, describing symptoms to an AI chatbot may appear harmless. So why are doctors concerned about people using it for diagnosis?
"Most patients who turn to Google or AI tools for information are already anxious about their symptoms," says Dr Vikas Mittal, director and pulmonologist at CK Birla Hospital, Delhi. "After reading multiple possibilities online, they often become even more anxious."
Dr Anupam Roy, additional director of nephrology and kidney transplant at Aakash Healthcare, says patients often arrive carrying "a mix of confusion, anxiety and uncertainty".
"I asked AI about my mouth ulcer and it gave me all the possible diseases in the world," says Anandeshwar.
Shraddha recounts a similar experience.
"It generally starts with the minor things, but it also gives something that can be pretty bad as well."
The opposite risk is excessive reassurance and validation.
"The diagnosis calms me down," Neha admits.
Manik further elaborates: "Chatbots are made in a way that gives you validation. Unless you context lock them in a different pattern, they aren't going to make you feel bad about anything."
Doctors are also aware of this pattern.
"AI can also provide generic reassurance because it lacks the context that would allow a clinician to identify subtle warning signs," says Dr Nikhil Rajvanshi, consultant in paediatric pulmonology at Madhukar Rainbow Children's Hospital.
"Breathlessness is a classic example," says Dr Mittal. "A patient entering 'I feel breathless' into an AI chatbot may receive numerous possible diagnoses. However, determining the actual cause requires a detailed history, assessment of associated symptoms, physical examination, and often diagnostic testing."
When AI Gets It Wrong
Experts say the risks become far greater when symptoms are serious or potentially life-threatening.
Dr Rajvanshi recalls a case where a child with a persistent cough was repeatedly reassured by online sources and AI-generated information that the problem was likely a lingering viral infection.
When eventually examined, the child was found to have an inhaled foreign body requiring bronchoscopy.
In another case, parents became convinced their child might have a rare lung disease after researching recurrent cough symptoms online. Clinical evaluation later revealed uncomplicated asthma.
"The online information increased anxiety without improving the diagnostic process," Rajvanshi says.
The Danger of Certainty
The biggest concern among clinicians is not misinformation itself but misplaced certainty. The issue extends beyond physical health to mental health as well.
Yukta Sharma, a trainee clinical psychologist, describes patients arriving with diagnoses they have effectively assigned to themselves.
One woman in her thirties arrived convinced she had ADHD and had already paid for an assessment before speaking to clinicians. Further evaluation, however, revealed that her difficulties stemmed largely from work-related stress, anxiety and resulting distractibility.
Another young woman arrived convinced she had schizophrenia after discussing her symptoms with AI. The symptoms she described, including social withdrawal and loss of interest in activities, were more consistent with depression than schizophrenia.
"The moment she came in with that belief, it became difficult to challenge it," Sharma says.
Clinicians' encounters with AI-diagnosed patients are remarkably consistent. Once a patient adopts an AI-generated explanation, clinicians often find it difficult to convince them that alternative interpretations may be more accurate.
Dr Roy says problems emerge when patients become attached to a specific diagnosis suggested online and are unwilling to consider other possibilities.
What AI Cannot See
Doctors repeatedly emphasise that diagnosis involves much more than matching symptoms to diseases. Dr Rajvanshi notes that what parents describe as wheezing could represent asthma, vocal cord dysfunction, airway obstruction or several other conditions.
The difference often becomes apparent only through examination, observation and detailed questioning.
"A physical consultation includes examination findings, observation, touch, smell, interaction and clinical judgment developed through years of experience," Rajvanshi says.
Many aspects of a physical examination remain impossible to replicate through an AI consultation.
Dr Roy points to physical signs like swelling, pulse characteristics and blood pressure variations that cannot be assessed through AI consultations.
The limitation extends to psychological assessment as well.
As Sharma points out, a chatbot cannot see whether a patient becomes distressed while discussing a particular topic, whether they hesitate before answering a question, or whether their body language contradicts their words.
"Patients communicate far more than words," Dr Rajvanshi says.
A Tool, Not a Replacement
Despite these concerns, experts are not entirely anti-AI and encourage its responsible use.
Dr Mittal says AI can help patients understand medical terminology and prepare questions before consultations. Dr Roy believes it can improve health literacy when used appropriately. Dr Rajvanshi advises patients to use AI to become informed rather than diagnosed.
People themselves largely seem to understand this distinction.
Shraddha says she almost always sees a doctor afterwards.
Neha uses AI to gauge whether something might be serious, but does not alter her behaviour based on the chatbot's advice.
Even Manik, perhaps the strongest advocate among those interviewed, acknowledges that severe problems require professional care.
"When the problem is severe, it does tell me to go see a doctor," he says. "And thanks to the consultation, I am always sure about the recovery."
The emerging middle ground is neither complete reliance on AI nor outright rejection of it, but a more informed and responsible use of the technology. As medical science increasingly integrates AI to strengthen patient care, people can use these tools to better understand their health and engage more meaningfully with their doctors.
