AI is driving an "expertise recession" as people increasingly question professional advice

[midian182]

In brief: The rise of generative AI continues to impact society in unexpected ways. According to a new poll, one of these is making people less willing to take the advice of professionals such as doctors without first confirming it with a chatbot.

It used to be that most people would blindly accept the advice of professionals without questioning it. But new research by Use.AI shows how doctors, teachers, managers, and others are being challenged by AI.

The survey found that just over half (53%) of the 15,428 respondents now use AI to check an explanation from a professional before deciding whether they agree with it.

The other headline result is that 46% said professional credentials now carry less weight than they once did because AI users can independently investigate claims that would previously have been difficult to assess.

AI also means that many people now prepare questions for doctors or other professionals ahead of their appointments, allowing them to ask about subjects they might not otherwise have thought of.

Use.AI has defined this change as an "expertise recession," a decline in automatic deference to professionals who have years of training, experience, and often familiarity with an individual's case.

The problem, of course, is that AI doesn't always get things right, and that can have severe consequences when it comes to medical queries. A chatbot is unlikely to know everything about a patient's medical history, test results, or other details a doctor would have access to.

Last year, a man developed the rare psychiatric disorder bromism after following ChatGPT's suggestion to replace dietary salt with sodium bromide, while a lawsuit alleges the chatbot recommended a dangerous drug combination that resulted in a teenager's fatal overdose. It was also found that Google's AI health summaries cited YouTube more often than authoritative medical sources, and research showed that even a tiny amount of false medical information in training data could significantly increase incorrect answers.

Interestingly, there is a clear age-based split in people's willingness to question professional advice. 70% of adults aged 18 to 34 said AI has made them more comfortable questioning professional advice, compared with 49% of those aged 45 and over.

Ultimately, being confident doesn't mean being right, which should be the disclaimer for all AIs. "AI has democratised confidence faster than it has democratized knowledge," said Ihor Herasymov, co-founder and CEO of Use.AI. "Someone can use AI to arrive at better questions without becoming qualified to judge the answers."

Permalink to story:

AI is driving an "expertise recession" as people increasingly question professional advice

 

[GeoffreyA]

There should be more scepticism applied to authority, and human slop brought under scrutiny too.

 

[Theinsanegamer]

AI is simply getting the blame for making the type of research that used to take weeks, or even months, and compresses it into a few minutes.

What's driving "expertise recession" is decades of "experts" showing they have no idea what they are talking about. They make grandiose, controversial, or conflicting claims, then nothing happens. And more recently, people are threatened with legal action if they disagree with "experts". You can only do that so much before your "expertise" is treated with all the authority of a crystal ball.

Just as the internet exposed the lies of the media industrial complex, so to shall AI expose charlatans that claim authority based on their knowledge, hypocritical and contradictory as it may be.

 

[RudyBob]

What's driving "expertise recession" is decades of "experts" showing they have no idea what they are talking about. They make grandiose, controversial, or conflicting claims, then nothing happens. And more recently, people are threatened with legal action if they disagree with "experts". You can only do that so much before your "expertise" is treated with all the authority of a crystal ball.

Yes!!

 

[pnntmp]

That's one of the many positive effects of AI that are usually overlooked (or deliberately ignored).
Encouraging people to acquire more information and then make their own judgement, instead of blindly accepting other opinions - that's the basis of the ever more uncommon critical thinking.

If AI at its current level were available 5-6 years ago, certain expert opinions would have caused far less damage.

Citing anecdotal cases of AI being wrong is manipulative ... experts are also quite often wrong. And quite often not available when needed. Fortunately the latter is never the case with AI, models are instantly available, and every information they provide can be instantly verified with other models.

 

[jonny888]

There should be more scepticism applied to authority, and human slop brought under scrutiny too.

That has, and always will, be true. Critical thinking is an underutilised skill for many people. Rarely should one person's claims be taken as fact without additional fact/cross checking.

What's driving "expertise recession" is decades of "experts" showing they have no idea what they are talking about. They make grandiose, controversial, or conflicting claims, then nothing happens. And more recently, people are threatened with legal action if they disagree with "experts". You can only do that so much before your "expertise" is treated with all the authority of a crystal ball.

Just as the internet exposed the lies of the media industrial complex, so to shall AI expose charlatans that claim authority based on their knowledge, hypocritical and contradictory as it may be.

On the flip side, if I need brain surgery and the doctor involved with 20+ years experience tells me I need to do X to prepare for it, but random.AI tells me something else, I'm not going to assume the doctor is a hack and that his experience counts for nothing. I might question them "Why not do Y instead?", but that's very different to a blanket dismissal of someone with years of domain knowledge and experience "because AI said so".

Or to put it another way: apply that same critical thinking to anything that is generated by an LLM that by it's very design and implementation is capable of hallucinating and producing incorrect results.

 

[godrilla]

This is more nuanced imo. New practioners and mid-level practioners lack the experience to make the best decision compared to tenured veterans that have the experience but are less warm to newer medicine. There is nothing wrong with using LLMs to make sure there is no blind spots. If you ask your Prescriber about the potential differential diagnosis ( via llm provides) and they just brush you off without explaining their position then you should get another professional opinion imo. When it comes to your own health you have to be your own advocate do your own research , and seldomly take a leap of faith on one human practitioner no matter the experience.

 

[godrilla]

AI is simply getting the blame for making the type of research that used to take weeks, or even months, and compresses it into a few minutes.

What's driving "expertise recession" is decades of "experts" showing they have no idea what they are talking about. They make grandiose, controversial, or conflicting claims, then nothing happens. And more recently, people are threatened with legal action if they disagree with "experts". You can only do that so much before your "expertise" is treated with all the authority of a crystal ball.

Just as the internet exposed the lies of the media industrial complex, so to shall AI expose charlatans that claim authority based on their knowledge, hypocritical and contradictory as it may be.

There is a new llm based on evidence called openevidence that uses tertiary resources, public studies, peir reviewed journals,
new guidelines which makes it less hallucinating. FYI.
Imagine being diagnosed wrong, taking the wrong medicine, getting unnecessary labs procedures, surgeries and paying for unnecessary medical bills and taking off valuable time from work/family via recovery or medical procedures that are unnecessary in 2026?

 

[arrowflash]

One thing that heavily concerns me is that many younger doctors are actually using LLMs themselves, both for quick research and even for prescriptions. I've seen it. All it takes is for a practitioner, especially an overworked doctor in the last hours of a long shift, to turn off their brain when doing research through AI (like many people do) and a hallucination when consulting for a prescription, for disastrous consequences.

 

[Rumiku]

Been doing this for almost 14 years now; sans AI. When I took my youngest daughter, at just 6 weeks old, to the ER, I was strong armed by doctors on invasive tests and care I did not agree with and indeed, turns out was incorrect steps of action right from the AAP's recommendation and second opinion recoiled in horror at the initial triage. Because I did not say "yes" to everything and challenged a few of the calls, CPS was called for investigation as well.

The fall of doctors god-complex certainly brings a smile to my face, AI or not. Even in emergency situations (thankfully rare), someone is researching and becoming familiar with necessary versus invasive, care. It's come in handy quite a few times.

 

[jonny888]

That's one of the many positive effects of AI that are usually overlooked (or deliberately ignored).
Encouraging people to acquire more information and then make their own judgement, instead of blindly accepting other opinions - that's the basis of the ever more uncommon critical thinking.

If AI at its current level were available 5-6 years ago, certain expert opinions would have caused far less damage.

Citing anecdotal cases of AI being wrong is manipulative ... experts are also quite often wrong. And quite often not available when needed. Fortunately the latter is never the case with AI, models are instantly available, and every information they provide can be instantly verified with other models.

1. The information people are "acquiring from AI" is coming from an opaquely-data-trained source that they have no way to verify the integrity of. That's not to say everything it says it wrong, but you have no way of knowing whether what it regurgitated was true, hallucinated, or false to begin with. It is NOT a good source of RELIABLE information.

2. You've been vague about which expert opinions you think would have been avoided, but it's also a strawman that's easily inverted. "If people were using AI to this degree 5-6 years ago, a lot of damage due to misinformation/hallucination would have been done". Neither is verifiable without a telescope into the alternate timelines. It's just noise.

3.
I) AI hallucinations are not anecdotal, they are statistically inevitable based on their design and how they work.
ii) Pointing out this fact is not manipulative, it's an important part of critically assessing how much trust should be placed in LLM output.
iii) Yes, people/experts can SOMETIMES be wrong, as-is human nature, and which is why fact checking and critical thinking are important. But LLMs, in addition to their chances of hallucination, are trained on data generated by said-humans, and are perfectly capable of regurgitating any bad information that was fed into them. So while they may be convenient, they are in no way more trustworthy or a substitute for proper research.

 

[J Oelschl]

These "experts" have no one but themselves to blame. I had one doctor that wanted to prescribe me a drug before the blood test even came back and my questions were dismissed with prejudice. Doctors are getting soft bribes from big pharma in the form of "paid for conferences" in vacation destinations. It seems every profession has been infected with self-serving individuals to the point where my trust is very low in a lot pf things. I use AI to investigate a subject not for the answers per se but rather to know some intelligent questions to ask professionals.

 

[wiyosaya]

My doctor continually ignores my history and as a result, makes inane, IMO, comments about my test results. Plus, the "health system" of which his practice is a part, has a motto born of Hubris - "Medicine of the Highest Order" that has infected that entire health system and makes many providers in that health system think they are supermen and superior doctors when in fact, they make mistakes, too. I'm currently looking for a different doctor because of that.

However, I would never trust anything I got from an LLM as a basis for any medical decision I make for myself. I would probably find it like my current Doctor only on steroids.

 

[godrilla]

My doctor continually ignores my history and as a result, makes inane, IMO, comments about my test results. Plus, the "health system" of which his practice is a part, has a motto born of Hubris - "Medicine of the Highest Order" that has infected that entire health system and makes many providers in that health system think they are supermen and superior doctors when in fact, they make mistakes, too. I'm currently looking for a different doctor because of that.

However, I would never trust anything I got from an LLM as a basis for any medical decision I make for myself. I would probably find it like my current Doctor only on steroids.

Yes never trust an llms at face value but confrim and consider the results as a possibility. If the llm finds the resources that you have a question for you can read it yourself to make up your own mind. Hallucinations should always be the default concern. If you hit a wall with human intervention and are still in distress or ailment it wouldn't hurt to have an llm to look for potential blind spots that were not considered. For example did you know that it is not routine for check for thyroid levels in pregnancy? Meanwhile some women feel lethargic while pregnant because their thyroid levels may be affected? Most practitioners treat the number not the patient.

 

[GeoffreyA]

That has, and always will, be true. Critical thinking is an underutilised skill for many people. Rarely should one person's claims be taken as fact without additional fact/cross checking.

Sadly, many ills in the world and in history have sprung from this lack of critical thinking.

Been doing this for almost 14 years now; sans AI. When I took my youngest daughter, at just 6 weeks old, to the ER, I was strong armed by doctors on invasive tests and care I did not agree with and indeed, turns out was incorrect steps of action right from the AAP's recommendation and second opinion recoiled in horror at the initial triage. Because I did not say "yes" to everything and challenged a few of the calls, CPS was called for investigation as well.

The fall of doctors god-complex certainly brings a smile to my face, AI or not. Even in emergency situations (thankfully rare), someone is researching and becoming familiar with necessary versus invasive, care. It's come in handy quite a few times.

With the way doctors often are, it is highly advisable to research (come prepared, ask them questions) and getting second opinions.

 

[Arkitekt78]

Enjoying the irony of a bunch of people that think they know everything applauding AI replacement of a bunch of people that think they know everything.

 

[jbright]

Using AI to help formulate better questions for your doctor seems genuinely useful. Using it as the final authority that decides your doctor is wrong seems considerably less brilliant.

 

[DAOWAce]

Good.

Medical malpractice is a massive problem in the US, and our laws aren't at a place where I'd consider them anywhere near protective enough. 'Malpractice' feels like a routine practice, and very few cases have doctors lose their license or have harsher sentences put on them. Since healthcare is barely affordable in the US (compared to free in other countries), most people are already in a poor financial state and likely have no path for lawsuits, and it's much harder to prove negligence (audio/video recording in rooms isn't a thing anywhere I've been). But I digress.

After developing a chronic condition late last year, I turned to AI to get answers to my questions early into the condition that no doctors so far have been able to provide (including a 5 month wait for neurologist who knew nothing of the condition and recommended a place out of state). I've been incredibly well informed now to the point that if a doctor tries telling me something, I usually already know what it's about, or if it's true or false in relation to my condition, or have the ability to much more easily verify their claims or look for alternatives (existed before with things like pubmed or google scholar, but is now infinitely easier since an AI aggregates everything and does work in minutes that would take you hours/days).

My healthcare company just added its own AI, called "PatientGPT", and advertised it right in the dashboard after login. It's an absolute POS and when I tried to give it a brief description of my problem to test it out, it said "this is a medical emergency, call 911" and disabled the chat. It's even stupider than Gemini, somehow. Something like this is NOT ready for the average joe.

Personally I've hated the healthcare industry my entire life and any time I went in for a procedure, I always came out worse (but not dead--yet--like Monty Oum). This chronic condition developed after one of them, further proving my distrust. But, because you need to actually get tests done for diagnosis and because certain drugs can actually help conditions more than they hurt your body (and they ALL hurt your body), you still have to deal with them to live a less bad life.

Plus, a doctor is just one person. Every single doctor has their own knowledge repository, their own way of doing things, etc.. Your doc could do their own personal research and be very knowledgeable in your condition, while another could know absolutely nothing about your condition, yet still have the same job field. Getting multiple opinions has been the norm for decades. An AI is thousands of opinions, though usually steered by what's most available in public medical literature.

If used well, AI is a great tool. If blindly trusted, well, you've still got the same problems. Manipulation to filter content is another big problem, which all of the main LLM's suffer from.

Also, AI is very useful to help summarize a health journal you're keeping and print out a concise sheet to bring to appointments. Unfortunately, most places don't want to read anything (especially hospitals).. yet another problem.

If anyone is wondering what my issue is, me too. I only have a blanket "dysautonomia" analysis so far with theories about the types and cause. What exactly caused it, what exact type it is, what exactly can be done to treat it, is all still up in the air.

 

[GrayDragon]

AI is simply getting the blame for making the type of research that used to take weeks, or even months, and compresses it into a few minutes.

What's driving "expertise recession" is decades of "experts" showing they have no idea what they are talking about. They make grandiose, controversial, or conflicting claims, then nothing happens. And more recently, people are threatened with legal action if they disagree with "experts". You can only do that so much before your "expertise" is treated with all the authority of a crystal ball.

Just as the internet exposed the lies of the media industrial complex, so to shall AI expose charlatans that claim authority based on their knowledge, hypocritical and contradictory as it may be.

Doubtful. I don't think the percentage of charlatans is large enough to support your conclusion. I believe most people are trying to give you good value for your payment for their services. Incidentally, many professionals also have to maintain certifications that, among other things, include a code of ethics. I've never heard of an AI that had any kind of code of ethics.

 

[GrayDragon]

Good.

Medical malpractice is a massive problem in the US, and our laws aren't at a place where I'd consider them anywhere near protective enough. 'Malpractice' feels like a routine practice, and very few cases have doctors lose their license or have harsher sentences put on them. Since healthcare is barely affordable in the US (compared to free in other countries), most people are already in a poor financial state and likely have no path for lawsuits, and it's much harder to prove negligence (audio/video recording in rooms isn't a thing anywhere I've been). But I digress.

After developing a chronic condition late last year, I turned to AI to get answers to my questions early into the condition that no doctors so far have been able to provide (including a 5 month wait for neurologist who knew nothing of the condition and recommended a place out of state). I've been incredibly well informed now to the point that if a doctor tries telling me something, I usually already know what it's about, or if it's true or false in relation to my condition, or have the ability to much more easily verify their claims or look for alternatives (existed before with things like pubmed or google scholar, but is now infinitely easier since an AI aggregates everything and does work in minutes that would take you hours/days).

My healthcare company just added its own AI, called "PatientGPT", and advertised it right in the dashboard after login. It's an absolute POS and when I tried to give it a brief description of my problem to test it out, it said "this is a medical emergency, call 911" and disabled the chat. It's even stupider than Gemini, somehow. Something like this is NOT ready for the average joe.

Personally I've hated the healthcare industry my entire life and any time I went in for a procedure, I always came out worse (but not dead--yet--like Monty Oum). This chronic condition developed after one of them, further proving my distrust. But, because you need to actually get tests done for diagnosis and because certain drugs can actually help conditions more than they hurt your body (and they ALL hurt your body), you still have to deal with them to live a less bad life.

Plus, a doctor is just one person. Every single doctor has their own knowledge repository, their own way of doing things, etc.. Your doc could do their own personal research and be very knowledgeable in your condition, while another could know absolutely nothing about your condition, yet still have the same job field. Getting multiple opinions has been the norm for decades. An AI is thousands of opinions, though usually steered by what's most available in public medical literature.

If used well, AI is a great tool. If blindly trusted, well, you've still got the same problems. Manipulation to filter content is another big problem, which all of the main LLM's suffer from.

Also, AI is very useful to help summarize a health journal you're keeping and print out a concise sheet to bring to appointments. Unfortunately, most places don't want to read anything (especially hospitals).. yet another problem.

If anyone is wondering what my issue is, me too. I only have a blanket "dysautonomia" analysis so far with theories about the types and cause. What exactly caused it, what exact type it is, what exactly can be done to treat it, is all still up in the air.

According to Google's AI, you mean that your "autonomic nervous system does not work properly." That sounds like quite a challenge but have you tried meditation? Regular practice of relaxation techniques seems like a good start to controlling what sounds like a psychological issue.