General Discussion

Ms. Toad

(38,351 posts)

4. From the number of people I've talked to using ChatGPT this way -

Mon Jan 5, 2026, 10:34 PM

Jan 5

I'm sure the numbers are accurate.

I'm in a lot of communities which are very heavy consumers of medical services. At least once a day, in one of these communities, I see a suggested to post the results of an MRI, or a pathology report, or something similar into ChatGPT for an interpretation.

No matter how often I explain that ChatGPT is designed to be conversational, not factual, they insist it is the same as doing a Google search.

For anyone who thinks it is the same - it's not - in a Google search, you get links to actual websites and can verify the reliability of the website as part of evaluating the information. Generally, those web pages link you to additional pages (whose reliability can be verified, etc.) In ChatGPT you get a word dump - with no clues as to where it got the information from - or if it just made it up out of thin air.

Edit history

Please sign in to view edit histories.

Recommendations

0 members have recommended this reply (displayed in chronological order):

14 replies

= new reply since forum marked as read

Highlight:

Money-losing OpenAI wants to make money from healthcare - wants the WORLD'S medical data, AND federal support [View all] highplainsdem Jan 5 OP

What ever happened JustAnotherGen Jan 5 #1

They see BigMin28 Jan 5 #2

They not only want federal tax dollars, they want governments to ensure they'll never have to pay highplainsdem Jan 5 #7

HIPAA SCHMIPPA purr-rat beauty Jan 5 #3

From the number of people I've talked to using ChatGPT this way - Ms. Toad Jan 5 #4

It's really scary that they're doing that. Musk has encouraged Twitter users to use Grok to interpret highplainsdem Jan 5 #5

I know - Ms. Toad Jan 5 #6

Do these people realize at all that chatbots can hallucinate? Would showing them video of highplainsdem Jan 6 #8

I tell them, every chance I get. Ms. Toad Jan 6 #9

I'm sorry to hear that, and I sympathize. Kudos for trying to help them. highplainsdem Jan 6 #10

Even though they aren't listening yet, I still want to point out some info in reply 11, from tweets I highplainsdem Jan 6 #13

The biggest point I try to get across is that running it through AI removes all clues you might follow Ms. Toad Jan 6 #14

1 in 5 AI diagnoses of cancer contain inaccurate information. A 23% error rate. highplainsdem Jan 6 #11

Kick dalton99a Jan 6 #12