MedPage Today | Medical News
Region: US
Friday 27 January 2023 20:52:55 GMT
2352555
158582
3872
33491
Music
Download
Comments
baaj :
OpenAI absolute confidence in its confabulations is actually dangerous. It often doesn’t back down when challenged
2023-01-27 22:31:33
9713
André Luís 🇵🇹🇪🇺 :
People don’t understand it’s only guessing the most statistically probable answer, not a factually correct answer. Don’t trust it.
2023-01-28 22:27:26
1990
dovnetanel :
Cause chatgpt is a chat ai it's priority is being conversational not correct, that's why we need search engine/ professional ai
2023-01-28 15:32:06
2830
Dasdutt :
So openAI is a Republican?
2023-01-27 23:02:46
2035
dream2denali :
We do not jump to PE because it is likely, we jump to PE because it is deadly…if I rule out a PE, I have time to work my way back to costochondritis.
2023-01-28 04:54:47
747
jared :
Everything it creates is made up. It’s a language model. It’s not lying it’s generating realistic content
2023-01-27 22:46:00
1698
Philippe Haussmann :
OpenAI isn’t ‘willing to lie’, it’s doing it’s best to produce real seeming conversation. It’s wrong but can’t be made to understand
2023-01-28 01:57:27
1932
Left Mike :
OpenAI doing anti-vaxxer level research 😂
2023-01-29 03:52:43
805
Christine | PharmD | Boston :
So openAI is another gaslighting doctor🧐
2023-01-30 03:44:36
1552
OG Process :
it's a text completion engine not an answer machine. important distinction.
2023-01-27 23:11:23
959
Dr. Christina | OrganizedMedic :
Not me confidently saying “PE” 😂😂
2023-01-29 15:21:32
30
Dr Kat, Epidemiologist :
I agree that it’s not good for providing evidence - that’s what I’m telling my students too
2023-01-28 19:39:20
145
SlowMotionIdea :
so openAI did not finish his/her residency yet.
2023-01-30 04:06:05
92
Isaac Taylor :
Listen dude—it's an NLP language model. It's about making something that sounds right. You need a different ML model to diagnosis.
2023-01-28 02:00:57
520
Austin Meyer :
This “lie” feels like a misunderstanding about what the model is doing. It is a generative model made to make up realistic-sounding text.
2023-01-28 03:32:15
283
Brandon :
Open AI doesn’t currently pull data from the internet. So it wouldn’t be able to give real sources. So it just makes up what they would look like.
2023-01-28 20:56:49
281
spitfir23 :
open ai does this a lot. I a Q that it said was true. When provided evidence proving otherwise (that it said) it blatantly said 4 > 20
2023-02-06 06:01:02
1
ReerSambuus :
That’s wild!!
2023-01-27 23:23:48
51
kreu327 :
This is just proof that some ppl don't know a thing about how this tech works, how to use it, or its current limits 🙄 that's the real danger.
2023-01-28 14:24:40
82
Bacchus 🌐 :
This is the thing. A language model doesn’t understand reality; it just puts together different words that seem likely to go together.
2023-01-28 22:49:44
39
mRNAyonnaise :
They say we build our own biases into the algorithms.
2023-01-28 04:53:34
84
not_carlos_pena :
Honestly just getting differentials and ranking them by a % likelihood would be great from an Ai.
2023-01-28 00:20:10
101
Suci :
I asked OpenAI questions on Dialogflow CX and it answered them perfectly until I realized those were for Dialgoflow ES. 😅 Almost got fooled
2023-01-28 05:01:29
17
To see more videos from user @medpagetoday, please go to the Tikwm
homepage.
