Cleo Abram
Region: US
Friday 31 May 2024 15:47:14 GMT
111263
7751
75
242
Music
Download
Comments
septemberpiano :
Problem is when it gets things wrong, it get them REALLY wrong. And tends to be unable to correct itself.
2024-06-01 07:52:45
239
Saf, the Artist :
Is Jennifer Garner your doppelgänger?
2024-06-02 04:06:20
47
exstraняшка [она/её] :
what AI parameters were used? or did they just use the chat.openai thingy?
2024-08-24 17:06:36
0
randomthings :
cool
2024-06-17 05:53:57
10
oshanshan248 :
for both humans and AI/ML, we still gotta question the internal validity. how well do these tests match doctor performance in practice in the real world?
2024-06-05 21:01:06
25
Josh Nagel :
this is incredible (early-stage) news! GPT is showing that it's a front runner when it comes to being a tool for more everyday and (potentially) every industry.
2024-05-31 15:59:13
55
I'm not telling you :
Doctors, apart from surgeons, mostly need to just recall information and put pieces together. This is something LLMs would be perfect at. And could lower the amount of medical mistakes.
2024-06-01 12:47:17
16
AAAAAGHSH :
basically it's not a very good test
2024-07-30 10:08:51
0
Leonthebleh :
Large language models will do that. Some of those terms are rather specific. So it would make sense that gpt4 would have come across something about the subjects. I think machine learning on imaging 1
2024-05-31 17:10:40
31
Walter Davis :
I am worried about AI and not understanding nuance and exceptions to the rule
2024-05-31 23:05:05
21
natvanvan403 :
Actual practice (of any profession) is completely different than a silly rote-memorized exam. We need to stop conflating the two.
2024-06-01 21:44:06
3
Andy Ma :
Why does the graph show Llama and Palm2 when the latest models are Llama3 and Gemini 1.5? And no Claude3? Is this an old study?
2024-05-31 23:35:35
13
Bill Huff :
Thank you. 😎
2024-06-01 18:37:08
0
Lux Evilian :
again, nice
2024-05-31 19:48:07
0
Cayle :
I think this is great actually! AI finally having positive usage rather than just feeling like it’s screwing over the artists/writers, but that’s also a different discussion, lol.
2024-05-31 17:08:03
0
Ski_Sk8_Brd :
Last year I tried out ChatGPT and asked how to do a salchow figure skating jump. It gave an incorrect answer 4 times. 1st answer wrong, 2-4 wrong after being told previous answer was incorrect.
2024-06-02 04:15:03
2
Andre :
As an engineer, I use the information from AI models extensively. One prerequisite is that one should approximately know the answer to determine the likely accuracy of the AI model.
2024-06-18 15:52:19
0
⚔ :
now imgn if they trained an ai on only doctor related things so it hallucinations are less
2024-06-01 13:41:39
1
Craig Penfold 🏴 :
scary
2024-06-02 16:36:40
0
Acker ༽ :
first
2024-05-31 15:53:01
0
mattbdev :
I think that a small language model may be better suited to specialize and help doctors.
2024-05-31 22:28:15
1
melissalem929 :
🥰🥰🥰
2024-07-12 21:16:07
3
lisaxxxxxxxxxx :
@Cleo Abram take @ashley with u next time u go discover nasa or any science stuff she's so funny 😂
2024-06-03 02:34:00
0
W :
💚💚💚
2024-05-31 18:07:44
0
To see more videos from user @cleoabram, please go to the Tikwm
homepage.
