News for hackers. Stuff that matters.

A comment on Study: ChatGPT Outperforms Physicians in Quality, Empathetic Answers to Patients. Parent comment

Re:Study: ChatGPT Outperforms Physicians in Quality, Empathe... by catlifeonmars on Saturday April 29, 2023 @12:30PM (#35754428)
LLMs currently have this problem where they will give confident-sounding responses to prompts where they are lacking enough context. Humans are built to read that as accuracy. It’s wholly a human interface problem.
[ Parent | on HN ]
Re:Study: ChatGPT Outperforms Physicians in Quality, Empathe... by rep_lodsb on Saturday April 29, 2023 @02:14PM (#35755532)
Some humans find that style of response infuriating, but apparently we are in the minority.

It's almost like "AI hacking people's brains" turned out to happen accidentally, and a huge number of supposedly smart people are getting turned into mindless enthusiasts by nothing more than computer generated bullshit.

[ Parent | on HN ]
Re:Study: ChatGPT Outperforms Physicians in Quality, Empathe... by kosh2 on Sunday April 30, 2023 @03:18AM (#35760338)
Agreed. It would be great if it could ask questions about missing context but maybe language models are bad for that task. Maybe one needs a second run, evaluate the answer and then make it look for info that could improve the answer.
[ Parent | on HN ]