Consumer behaviour study
AI Before a Doctor? Here's Why
In a Hercules survey of 400 urban Indian adults, 30.3% would look to an AI assistant such as ChatGPT first for a new symptom and 28.5% would go to a doctor. 98.3% have asked AI about a symptom, and 76.8% trust AI health information completely or somewhat.
Explore the findingsPick an AI assistant such as ChatGPT first for a new symptom
Pick a doctor or healthcare professional first
Have ever asked an AI assistant about a symptom
Based on 400 completed responses. Consumer behaviour study.
What we set out to answer, and what we found
We tested where urban Indians look first when a new symptom appears
Where do people look first?
We asked which single source people are most likely to check first for a new health symptom, and which sources they have used before speaking with a doctor.
How widely is AI tried and used?
We measured whether people have ever asked an AI assistant about a symptom and how often they use AI for health questions.
How far is AI trusted?
We asked how much people trust AI-generated health information.
How comfortable is AI as a first source?
We asked for a 0 to 10 comfort rating about using AI as the first source for a new symptom, in the respondent's own words.
AI and doctors are neck and neck as the first stop. 30.3% of 400 respondents would check an AI assistant first for a new symptom and 28.5% a doctor. The 2-point lead is not statistically different. Together they take 58.8% of first looks.
Search is third, and everything else is small. Google or a search engine is first for 19.8%, health websites for 8.8%, YouTube or social media for 7.2% and family or friends for 4.5%.
Almost everyone has tried AI for symptoms. 98.3% have asked an AI assistant about a health symptom. Among those who pick AI first, 39.7% use it often and 24.8% very often.
Trust is high and distrust is rare. 57.3% trust AI health information somewhat and 19.5% completely, against 1.8% who distrust it somewhat. 21.5% sit in the middle.
Comfort rests on AI as a starting point. Of the 145 coded comfort answers, 76.6% are a bare top rating, 21.4% describe AI as an educational starting point and 2.1% give a strong rating with a reason.
Distrust AI Health Info
Distrust it somewhat, against 76.8% who trust it somewhat or completely
30.3% check an AI assistant first for a new symptom, 28.5% a doctor
We asked everyone in the sample which single source they are most likely to look at first when they notice a new health symptom.
Where Respondents Look First for a New Symptom
Single answer, % of respondents. Base: 400
View the data behind this chart
| Answer | Share |
|---|---|
| AI assistant such as ChatGPT | 30.3% |
| Doctor or healthcare professional | 28.5% |
| Google or search engine | 19.8% |
| Health website or app | 8.8% |
| YouTube or social media | 7.2% |
| Family or friends | 4.5% |
| Pharmacy or pharmacist | 0.7% |
| I usually don't look it up | 0.2% |
01. The AI and doctor lead is 2 points, and the report found no tested difference, so the two read as level.
02. The top two together reach 58.8% of everyone asked.
03. The report notes that age bands are similar on first stop, so one plan can serve every age group.
04. The data does not show which source people use next, or whether AI use comes before or after a doctor visit.
Hospital websites lead on sources used, with AI assistants fourth at 26.8%
In the screener, respondents picked every source they have used to understand a symptom before speaking with a doctor. People who selected None of these were screened out.
Sources Used to Understand a Symptom Before Seeing a Doctor
Multiple answers allowed, % of respondents
View the data behind this chart
| Answer | Share |
|---|---|
| Hospital or medical websites | 46.5% |
| Google or search engines | 40% |
| YouTube | 37.5% |
| AI assistants | 26.8% |
| Instagram or social media | 22.3% |
| Online health communities | 21% |
| Family or friends | 13.8% |
| Pharmacist | 9.3% |
01. AI assistants are used by 26.8% as one of several sources, yet they are the single most likely first stop for 30.3%.
02. Respondents use many sources before a doctor, with the top three each above 37%.
98.3% have asked AI about a symptom, and most AI-first people use it often
We asked whether people have ever asked an AI assistant about a health symptom and, for each first-stop group, how often they use AI for health questions.
How Often They Use AI for Health Questions, by First Stop
% of each first-stop group
View the data behind this chart
| AI assistant first | Doctor first | |
|---|---|---|
| Never | 0.8% | 0% |
| Rarely | 4.1% | 5.3% |
| Sometimes | 30.6% | 28.9% |
| Often | 39.7% | 28.1% |
| Very often | 24.8% | 37.7% |
Have Ever Asked an AI Assistant About a Health Symptom
% of respondents. Base: 400
View the data behind this chart
| Answer | Share |
|---|---|
| Male | 98.4% |
| Female | 97.4% |
| All respondents | 98.3% |
01. Among people who pick AI first, 39.7% use it often and 24.8% very often. Only 4.9% use it rarely or never.
02. Doctor-first people use AI too: 37.7% very often and 28.1% often.
03. Trial is 98.4% for men and 97.4% for women. The gap is untested, and the report does not read it as a gender difference.
04. Trial by age band and by city was not split out in the report, so it cannot guide where to invest.
76.8% trust AI health information, and only 1.8% distrust it
We asked all 400 respondents how much they trust AI-generated health information.
Trust in AI-Generated Health Information
% of respondents. Base: 400
View the data behind this chart
| Answer | Share |
|---|---|
| Trust somewhat | 57.3% |
| Neither trust nor distrust | 21.5% |
| Trust completely | 19.5% |
| Distrust somewhat | 1.8% |
How Often They Use AI for Health, by Level of Trust
% of each trust group. Base: 393 (trust somewhat n=229, neither n=86, trust completely n=78)
View the data behind this chart
| Trust somewhat | Neither trust nor distrust | Trust completely | |
|---|---|---|---|
| Often | 46.7% | 23.3% | 15.4% |
| Sometimes | 29.3% | 55.8% | 17.9% |
| Very often | 20.1% | 11.6% | 62.8% |
| Rarely | 3.5% | 9.3% | 2.6% |
| Never | 0.4% | 0% | 1.3% |
01. 57.3% trust AI health information somewhat and 19.5% completely. The 21.5% in the middle are the group to win over.
02. 62.8% of those who trust AI completely use it very often, against 20.1% of those who trust it somewhat and 11.6% of the neutral group.
03. The report does not show that trust changes how often people use AI, so frequency alone should not justify trust-building features.
04. Trust is 86.5% among those aged 31 and over, 75.3% for 18 to 21, 73.5% for 22 to 23 and 70.9% for 24 to 30. The report finds no tested age gap. Men are at 77.0% and women at 75.6%, and the female base is small.
Comfort with AI as a first source rests on it being a starting point
We asked for a 0 to 10 comfort rating about using AI as the first source for a new symptom. The question was answered in free text, and 145 answers were coded into themes.
Coded Themes in Comfort Answers
% of coded responses, multiple themes allowed. Base: 145
View the data behind this chart
| Answer | Share |
|---|---|
| Maximum comfort, bare rating | 76.6% |
| Comfort as educational starting point | 21.4% |
| Strong comfort with stated reason | 2.1% |
01. Most coded answers are just a high number, so they say little about why. Only 2.1% pair a strong rating with a reason.
02. 21.4% of coded answers describe AI as an educational starting point, and some of those answers say it should be paired with a healthcare professional.
03. Because the scale was captured as free text, the report says the numeric form should be confirmed before comfort scores are read as ratings.
“7 out of 10. AI is fast for understanding possibilities, but it should be paired with a healthcare p” Survey respondent, comfort answer (as printed in the report; coded as educational starting point)
“8 for acting as an educational starting point” Survey respondent, comfort answer (coded as educational starting point)
“9 I am very comfortable” Survey respondent, comfort answer (coded as strong comfort with stated reason)
What we would recommend
Practical priorities drawn from this study for further validation.
Show AI and doctor options side by side on the first-stop screen.
30.3% choose AI first and 28.5% a doctor. A product that offers only one will miss close to half of first looks.
Build a clear route from every AI answer to a doctor.
76.8% trust AI health information, and comfort rests on AI as a starting point. An escalation path matches how respondents describe their own use.
Launch AI symptom checking to all users, not a narrow group.
98.3% have already asked AI about a symptom, and the report finds no tested gender or age gap in trial or trust.
Focus trust work on the neutral 21.5%.
Only 1.8% distrust AI health information. The larger prize is the middle group, whose frequent use is just 11.6% against 62.8% for people who trust it completely.
Test what people do after the first stop before placing features.
The study does not show which source people use next, or whether AI comes before or after a doctor visit.