Should AI generated posts be banned or otherwise regulated?


I just wonder. 

At least, when I start a new thread, I am expecting other people's opinions.  I can get my own AI response so I am not sure why others would repeat what I can do myself. 

If someone were to have access to some better AI than I have access to, I guess that would be useful info I could not otherwise get.  But in general, I wonder why posters think responding with AI content is useful to someone who can get that directly themselves. 

jji666

In my experience, AI has been mostly spot-on related to the sonic signature of audio equipment that I own, have owned, or have heard and am considering trying in my system.  However, like other things, it is simply a tool. I wouldn’t use any one platform to base buying decisions on, including audio reviews, manufacturer information, or these forums where much of what gets posted is a parade of whatever gear or cables that others own.  Occasionally, when my knowledge base or fact checks indicate that the AI answer is wrong, I call it out and receive a response like, "yes, you are correct..." and then a discussion including the correct answer.  If anything, I feel a little guilty using AI because of all the computing power and associated energy it sucks up.  I believe we are in a phase where promoters are trying to generate interest and new users before starting to somehow charge for it.  

80% of medical AI queries are factually wrong.

Please verify the validity of this statement and provide reliable sources for this disclaimer.

Some self-identified AGRs (AudioGoneRs) tend to scrutinize AI critically, but may apply less consistent standards when evaluating their own community’s claims, sometimes making assertions without clear justification.

Law Firms have fired lawyers who don’t verify the source of AI sites.

Lawyers may be fired for providing false evidence or malpractice, regardless of whether the source is AI or not. One cannot testify in court using AI-generated evidence. Simulated crashworthiness cannot be presented in court as evidence for crash investigations.

To be clear, I am not against AI use.  I am probably "medium" when it comes to AI use...I use it daily but not like some people I know. IMHO AI can also be over- and mis- used. 

To refine my question, what I am really getting at is the type of post where it is simply and only "Here is what ChatGPT says:" without any actual thoughts from the poster.  That's just running a search for someone. 

My point, to perhaps put it better, is that each post should have some human element or that is contrary to the point of a forum. 

"80% of medical AI queries are factually wrong."

I assume this was a human generated answer. 

The claim is decent but requires significant qualification — it’s accurate in a narrow sense that gets badly distorted when stated baldly. 

A recent JAMA Network Open study evaluated 21 large language models across 29 clinical case scenarios and found that AI systems failed to provide the correct diagnosis in over 80% of early-stage cases, where symptoms are often non-specific.  The 80% figure applies specifically to early-stage differential diagnosis with minimal clinical information, not to medical queries generally.

Source:
 https://www.business-standard.com/health/ai-chatbots-misdiagnose-early-cases-80-percent-study-126041400329_1.html

But there’s more to it.

"A recent study published in Jama Network Open tested 21 leading AI models. The researchers asked them to work out possible medical diagnoses. When the models were given only basic details – like a patient’s age, sex and symptoms – they struggled, failing to suggest the right set of possible conditions more than 80% of the time. Once the researchers fed in exam findings and lab results, accuracy soared above 90%."

Source: 
https://theconversation.com/half-of-ai-health-answers-are-wrong-even-though-they-sound-convincing-new-study-280512

If precision is what you’re for, then be the change you want to see in the world.

John Dewey liked to say, "A problem well-put is a problem half solved."

If the question is not well put, the answer likely won’t be. Put in a better effort on your question/prompt and the answers will get better.

That does NOT mean that AI should take the place of human experience or that it’s worth the energy cost or that it’s valid to use for plagiarism, etc. But the claim being advanced is that it cannot  do a good job with facts. That just ain’t so in many cases.