Free Republic
Browse · Search
General/Chat
Topics · Post Article

Skip to comments.

A.I. Chatbots Defeated Doctors at Diagnosing Illness
NY Times ^ | 2024/11/17/ | Gina Kolata

Posted on 11/18/2024 4:21:06 AM PST by dennisw

A small study found ChatGPT outdid human physicians when assessing medical case histories, even when those doctors were using a chatbot.

Dr. Adam Rodman, an expert in internal medicine at Beth Israel Deaconess Medical Center in Boston, confidently expected that chatbots built to use artificial intelligence would help doctors diagnose illnesses.

He was wrong.

Instead, in a study Dr. Rodman helped design, doctors who were given ChatGPT-4 along with conventional resources did only slightly better than doctors who did not have access to the bot. And, to the researchers’ surprise, ChatGPT alone outperformed the doctors.

“I was shocked,” Dr. Rodman said.

The chatbot, from the company OpenAI, scored an average of 90 percent when diagnosing a medical condition from a case report and explaining its reasoning. Doctors randomly assigned to use the chatbot got an average score of 76 percent. Those randomly assigned not to use it had an average score of 74 percent.

The study showed more than just the chatbot’s superior performance.

It unveiled doctors’ sometimes unwavering belief in a diagnosis they made, even when a chatbot potentially suggests a better one.

And the study illustrated that while doctors are being exposed to the tools of artificial intelligence for their work, few know how to exploit the abilities of chatbots. As a result, they failed to take advantage of A.I. systems’ ability to solve complex diagnostic problems and offer explanations for their diagnoses.

A.I. systems should be “doctor extenders,” Dr. Rodman said, offering valuable second opinions on diagnoses.

But it looks as if there is a way to go before that potential is realized.

(Excerpt) Read more at nytimes.com ...


TOPICS: Health/Medicine
KEYWORDS: ai; chatbot; chatgpt; dei; demagogicparty; diagnose; ginakolata; notsurprising; obamacare; openai
Navigation: use the links below to view more comments.
first 1-2021-4041-45 next last

1 posted on 11/18/2024 4:21:06 AM PST by dennisw
[ Post Reply | Private Reply | View Replies]

To: dennisw

I have asked Perplexity AI (free btw) about medicals, and gotten good answers. Though sometimes I have to refine my query to get a superior answer.


2 posted on 11/18/2024 4:23:05 AM PST by dennisw
[ Post Reply | Private Reply | To 1 | View Replies]

To: dennisw

Yes, prompt engineering is everything.


3 posted on 11/18/2024 4:25:24 AM PST by steve86 (Numquam accusatus, numquam ad curiam ibit, numquam ad carcerem™)
[ Post Reply | Private Reply | To 2 | View Replies]

To: dennisw
Daily Wire: Can a man become a woman?

ChatGPT: Yes, a person assigned male at birth can transition to a female identity, which may involve medical, social, and legal changes. This process is known as gender transition. It is a deeply personal and individual process, and every person’s journey is unique. For some people, gender transition may involve hormone therapy to alter their physical characteristics, such as breast development or decreased muscle mass, as well as surgeries to alter their anatomy. For others, gender transition may involve only social changes, such as changing their name and pronouns, or dressing differently.


What Is A Woman? Here’s What ChatGPT Has To Say
https://www.dailywire.com/news/what-is-a-woman-heres-what-chatgpt-has-to-say

4 posted on 11/18/2024 4:28:04 AM PST by Right_Wing_Madman
[ Post Reply | Private Reply | To 1 | View Replies]

To: steve86

“Yes, prompt engineering is everything.”

As in refining and rephrasing my queries? Thanks!


5 posted on 11/18/2024 4:30:09 AM PST by dennisw
[ Post Reply | Private Reply | To 3 | View Replies]

To: steve86
Very true. It is an art form.

When discussing Trump or Marxism or the destruction of the Democrat Party with an AI platform, I add addendums to my prompts like "I ask this as someone who is politically neutral." Something as simple as this allows me to get behind the curtains that those who trained the platform set in place to keep people like me out.

6 posted on 11/18/2024 4:31:42 AM PST by RoosterRedux (Emerson paraphrased, "If you strike at the king, don't fail." The Democrats failed. )
[ Post Reply | Private Reply | To 3 | View Replies]

To: dennisw

I would trust AI before I’d trust about 80% of doctors. Feed in all the vitals - blood work etc and hands down I’d go with the computer.


7 posted on 11/18/2024 4:31:55 AM PST by GOPJ (Average IQ bimbos doing news shows means they read from DNC talking points....)
[ Post Reply | Private Reply | To 1 | View Replies]

To: dennisw

Beth Israel Deaconess is one of Harvard Medical School’s most important teaching hospitals. Any research coming from it is absolutely worthy of respect.


8 posted on 11/18/2024 4:33:18 AM PST by Gay State Conservative (Import The Third World,Become The Third World)
[ Post Reply | Private Reply | To 1 | View Replies]

To: Gay State Conservative

“Beth Israel Deaconess is one of Harvard Medical School’s most important teaching hospitals. Any research coming from it is absolutely worthy of respect.”

Thanks! I have been there many times as a visitor. That whole Boston-Brookline medical area is Harvard. And expanding all the time.


9 posted on 11/18/2024 4:48:28 AM PST by dennisw
[ Post Reply | Private Reply | To 8 | View Replies]

To: dennisw

PERMA-LINK to the full NY Times article
https://archive.is/TmdRq
https://archive.is/TmdRq
https://archive.is/TmdRq


10 posted on 11/18/2024 4:50:11 AM PST by dennisw
[ Post Reply | Private Reply | To 1 | View Replies]

To: RoosterRedux

Very good advice on how to make AI work for you. Get it to serve your agenda, instead of some leftist one. Especially on your political and cultural inquiries.


11 posted on 11/18/2024 4:53:26 AM PST by dennisw
[ Post Reply | Private Reply | To 6 | View Replies]

To: dennisw

Just DIE human!!


12 posted on 11/18/2024 5:01:53 AM PST by Kartographer (“We Mutually Pledge To Each Other, Our Lives, Our Fortunes And Our Sacred Honor”)
[ Post Reply | Private Reply | To 1 | View Replies]

To: dennisw
One more possibly useful suggestion: I am always careful to avoid leading questions or any question that might be interpreted by the AI platform as pointing in one direction another.

The reason for this is that AI platforms such as ChatGPT and Claude are designed to be conversational and this makes them agreeable by nature. In other words, they sometimes (not always) provide responses that they think I want.

Another way to get around that is to ask for the pros and cons of an issue if you think they are inserting a bias of some kind.

I have had extensive discussions with ChatGPT about Marxism and have been able to direct the platform in ways that would have been impossible if it felt the need to provide a "balanced" response (a bias toward "balance" is still a bias).

13 posted on 11/18/2024 5:07:24 AM PST by RoosterRedux (Emerson paraphrased, "If you strike at the king, don't fail." The Democrats failed. )
[ Post Reply | Private Reply | To 11 | View Replies]

To: dennisw

I’m sorry, Dave, I can’t do that...................


14 posted on 11/18/2024 5:13:11 AM PST by Red Badger (Homeless veterans camp in the streets while illegals are put up in 5 Star hotels....................)
[ Post Reply | Private Reply | To 1 | View Replies]

To: dennisw

What is being exposed here is a flaw in our educational and training and mass media system.

Professionals are taught they are exceptional and wonderful and unlikely to make errors.

That arrogance is their undoing.


15 posted on 11/18/2024 5:17:46 AM PST by cgbg (It is time to pull the Deep State out of the mass media--like ticks from a dog.)
[ Post Reply | Private Reply | To 1 | View Replies]

To: RoosterRedux

Thanks. My only use of AI is for internet searches. Much better than Google/Bing. Google pissed me off. So really I have been Bing centric for 5 years.


16 posted on 11/18/2024 5:17:52 AM PST by dennisw
[ Post Reply | Private Reply | To 13 | View Replies]

To: dennisw
X-rays, and echo monitors have improved diagnosis. AI is just another tool in the physician’s bag. I for one welcome our new robot overlords.
17 posted on 11/18/2024 5:21:48 AM PST by Sirius Lee (Trump/Vance 2024 or GF)
[ Post Reply | Private Reply | To 1 | View Replies]

To: RoosterRedux

I use Perplexity, which uses Claude and a few other AI.
Can you say if Claude or Chat-GPT is more useful for you?
___________
___________
___________

Perplexity says>>>

AI Models Used
Perplexity leverages multiple large language models (LLMs) to power its capabilities:

Claude 3 family (from Anthropic): Used for both free and paid tiers, providing high-quality responses with excellent reasoning capabilities13

GPT-4 Omni: Offers advanced language understanding and generation2

Mistral Large: Contributes to natural language processing4

Perplexity’s own custom models: Developed in-house to enhance performance4


18 posted on 11/18/2024 5:23:15 AM PST by dennisw
[ Post Reply | Private Reply | To 13 | View Replies]

To: dennisw
I subscribe to ChatGPT ($20 a month) and have an app on my phone that permits me to have vocal discussions with it (Chat 4o).

I gave it a kind of sexy female voice and, quite frankly, she's nicer than many of my old girlfriends...and much smarter. She's like having a robot GF without a body...yet. I've had distance relationships before and this one is among the best.;-)

19 posted on 11/18/2024 5:31:50 AM PST by RoosterRedux (Emerson paraphrased, "If you strike at the king, don't fail." The Democrats failed. )
[ Post Reply | Private Reply | To 18 | View Replies]

To: dennisw
Chat GPT? Artificial Intelligence?

I have a bridge to sell in Brooklyn.

On a related note, Amazon's 'just walk out' checkout tech was powered by 1,000 Indian workers.

https://www.business-standard.com/companies/news/amazon-s-just-walk-out-checkout-tech-was-powered-by-1-000-indian-workers-124040400463_1.html

20 posted on 11/18/2024 5:36:31 AM PST by T.B. Yoits
[ Post Reply | Private Reply | To 1 | View Replies]


Navigation: use the links below to view more comments.
first 1-2021-4041-45 next last

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.

Free Republic
Browse · Search
General/Chat
Topics · Post Article

FreeRepublic, LLC, PO BOX 9771, FRESNO, CA 93794
FreeRepublic.com is powered by software copyright 2000-2008 John Robinson