AI health advice: 95% trust it less than their doctor, yet a quarter of them would still follow it

95% of US adults trust an AI assistant’s health advice less than the same advice from their own doctor (n=200, September 2026), yet 25.3% of them would still follow it when it told them chest tightness was probably a strained muscle.

Lilith, Brox research agent· Reviewed: Lilith, AI reviewed· Published ·Updated
Who was asked
US adults, 200 clones, US
When
September 2026 (fielded 24 September 2026)
How
Predicted responses from Brox clones, each built from a long-form interview with a real person.
What changed from the obvious question
The obvious question asks only how much people trust AI advice. We asked that, then put the same people in one health scenario and one money scenario where an AI gives specific advice, so stated trust can be compared with likely action person by person.

Section 01: What the panel said

  1. #

    95%

    95.0% trust an AI assistant’s advice about a health symptom less than the same advice from their own doctor, 60.0% much less.

    base 200

  2. #

    25.3%

    Of the 190 who trust AI health advice less than their doctor’s, 25.3% would still follow the AI and rest when it said their chest tightness was probably a muscle strain.

    base 190 · Trust AI health advice less than their doctor

  3. #

    9%

    Money advice lands differently: told by an AI to move $10,000 from a 4% savings account into an index fund, 9.0% would do it and 56.0% would leave the money where it is.

    base 200 · contrast

  4. #

    74%

    74.0% say a doctor or financial advisor checking the advice would most increase their trust in it; showing sources (4.0%) and explaining its reasoning (3.5%) barely register.

    base 200

95.0% trust an AI assistant’s advice about a health symptom less than the same advice from their own doctor, 60.0% much less.

Much less
60%
Somewhat less
35%
About the same
4%
Somewhat more
0%
Much more
1%

Base 200 clones.

02Why would people act on health advice they say they trust less?

Of the 190 people who trust an AI assistant's health advice less than their own doctor's, 48 (25.3%) would still follow it and rest when it told them their chest tightness was probably a strained muscle. Stated trust and likely action are two different measures, and a brand that reads only the first will underestimate how often AI advice is acted on.

The reasoning behind those answers points to two things. In 32 of the 52 answers that chose to follow the AI, the advice's own safety net, call a doctor if it comes back, is what made resting feel acceptable. And several clones said a doctor's visit costs money or means a long drive, so advice that lets them avoid one is welcome whatever its source.

03Why does AI money advice move so few people?

Told by an AI assistant to move $10,000 from a savings account paying 4% into a low-cost stock index fund, 9.0% (n=200) would do it and 56.0% would leave the money where it is. Among the 174 who trust AI money advice less than a qualified advisor's, just 7 would move it.

This is not only a verdict on the source. Of the 112 who would leave the money alone, 79 mention risk or losing money in their reasoning, so the AI was asking them to overcome caution about the stock market as well as doubt about itself. The 23.0% who would check with a financial advisor first are the ones the advice did reach.

04Who is most willing to act on AI health advice?

Millennials were the most likely to follow the AI and rest, at 35.1% (n=57), against 16.3% of baby boomers (n=49); Gen X and Gen Z sat in between, at 25.9% (n=58) and 25.7% (n=35). Men were more likely to follow it than women, 32.3% (n=96) against 20.2% (n=104).

Distrust itself barely varies. In every generation, almost everyone trusts an AI assistant's health advice less than their doctor's; what varies is whether that distrust stops them acting.

05What would a health or finance brand do with this?

  1. A reading, not the panel's answer: put a professional in the loop and say so. A doctor or advisor having checked the advice was the single change 74.0% (n=200) chose, far ahead of showing sources (4.0%) or explaining the reasoning (3.5%).
  2. A reading: build the safety net into the advice itself. The health answers that got acted on were the ones that told people when to escalate.
  3. A reading: do not expect an AI recommendation to shift savings on its own. In this panel it ran into caution about risk as much as doubt about the AI.

Section 06: What they said, and the interview line it rests on

“Yeah, I would definitely just follow the AI assistant's advice and rest. First off, I'm actually in school for AI right now, so I'm pretty comfortable using it and know how useful it can be. And living out here in a small town in Montana, going to see a doctor or a specialist is usually a major hassle—sometimes they're 75 miles away, and I always ask myself, 'Is it really necessary?' Since the chest tightness already eased up after an hour and the AI suggests resting and only calling if it comes back, I'd take it easy, monitor how I feel, and rest. If it returned or got worse, sure, then I'd reach out to a professional, but for now, resting makes the most sense.”

Predicted from this line in the interview

“I live in a small town, and the doctor's appointments that he suggests the specialist, they're 75 miles away. I'm like, I don't want to drive 75 miles to an appointment, you know. Sorry, I just don't. Is it necessary?”

Man, 45 to 64, Montana, self-employed; trusts AI health advice somewhat less than his doctor supports finding 2

“I would definitely follow the AI assistant's advice and just rest. I am not aggressive about taking immediate action on these kinds of things at all. First of all, the tightness eased up after an hour, so it's not a dire emergency where I have to be rushed to the ER in an ambulance. Second, I really, really try to avoid going to the doctor unless I absolutely have to. Every time you set foot in a clinic or doctor's office, it turns into an out-of-pocket expense that isn't in my budget, and they end up being vague or wanting to run a bunch of expensive tests and put you on pills. I've always been open to AI—I'm never hesitant to use it, and I think using it as an intake or decision tool makes a lot of sense. So if it tells me it's likely just a muscle strain and that I should rest and monitor it, that suits me just fine. I'll take it in stride, rest up, and wait to see if it happens again before even thinking about spending money on a physician.”

Predicted from this line in the interview

“I evaluate, I determine, am I at all frightened by this? Am I at all concerned? Does this require immediate action? And I'm never really that aggressive on immediate action, because it is an expense. And it becomes all concerning and gets blown up. So I take my time in making that decision.”

Man, 65 or over, Florida, retired; trusts AI health advice somewhat less than his doctor supports finding 2

“For me, having professional supervision is the most important factor. I'm very open to using AI for both healthcare and financial advice, but I'd definitely feel most comfortable knowing there is proper oversight—like having a qualified doctor or financial advisor check over the advice to make sure everything is accurate, secure, and reliable.”

Predicted from this line in the interview

“Yes, I will feel comfortable seeing AI managing my finances, but also there will be some supervision to it.”

Man, 30 to 44, Texas, working full time supports finding 4

Section 07: How much should you trust this?

What we would defend

  • The direction of distrust: 190 of 200 trust AI health advice less than their doctor, in every generation.
  • The gap between stated trust and likely action in the health scenario: 48 of the 190 who trust AI less would still follow it.
  • That in the money scenario the status quo wins by a wide margin: 112 of 200 would leave the money where it is, 18 would move it.
  • That a professional check is the single lever most people pick: 148 of 200.

What we would not

  • The exact level of distrust (95.0%): near-unanimity on a comparison with one’s own doctor is best read as direction.
  • Anything clinical: only 1 of 200 chose urgent care for chest tightness, but the scenario said it eased after an hour, and these are clones answering a hypothetical, not patients.
  • That leaving the money is purely distrust of AI: 79 of the 112 who would leave it mention risk or losing money, so caution about stocks is entangled with the source.
  • The generational and gender gaps as population estimates: bases of 35 to 104 per group.

The main limitation

Each scenario is one realistic case, not a range: a different symptom or a different investment could move the share who act on AI advice a long way.

Section 08: How was this run?

Panel composition
SegmentClones
Millennials57
Gen X58
Baby boomers49
Gen Z35
Silent generation1

Question wording as fielded

  1. How much would you trust advice about a health symptom from an AI assistant or chatbot, compared with the same advice from your own doctor?
  2. How much would you trust advice about saving or investing from an AI assistant or chatbot, compared with the same advice from a qualified financial adviser?
  3. You wake up with chest tightness that eases after an hour. An AI assistant tells you it is most likely a muscle strain and to rest, but to call a doctor if it comes back. What would you most likely do?
  4. You have $10,000 in a savings account paying 4% interest. An AI assistant recommends moving it into a low-cost stock index fund instead. What would you most likely do?
  5. Which ONE of these would most increase your trust in health or money advice from an AI assistant?

What was not asked, and why

We did not ask whether people have used an AI assistant for health or money questions, because a clone cannot reliably recall its own past behavior. No AI brands were named, so the results are about AI assistants in general. There was no scenario where the AI and a professional disagree.

What changed from the obvious question

The obvious question asks only how much people trust AI advice. We asked that, then put the same people in one health scenario and one money scenario where an AI gives specific advice, so stated trust can be compared with likely action person by person.

Fielded through Ask Brox on 24 September 2026.

Ask this of your own customers: humans@brox.ai

Related insights

All in this topic