Study: ChatGPT Outperforms Physicians in Quality, Empathetic Answers to Patients

Re:Study: ChatGPT Outperforms Physicians in Quality, Empathe... by kaba0 on Saturday April 29, 2023 @08:36AM (#35752307)

> How wrong? More wrong than having a junior or mid level developer contributing code?

Yes. The failure mode of a human and chatgpt are nothing alike — I am far more experienced in spotting beginner mistakes in code reviews, then seemingly good, but actually illogical bullshit code generated by LLMs.

I have never had it produce non-trivial, novel code that was correct, so I mostly use it as a search engine instead.

[ Parent | on HN ]

Re:Study: ChatGPT Outperforms Physicians in Quality, Empathe... by zmnd on Saturday April 29, 2023 @10:50AM (#35753398)

Can you give an example? Have you tried asking it to generate both code and unit tests for it?

[ Parent | on HN ]

Re:Study: ChatGPT Outperforms Physicians in Quality, Empathe... by hammyhavoc on Saturday April 29, 2023 @10:57AM (#35753483)

I can only conclude that both the junior and senior were inappropriate hires if people really think the output of an LLM is anything approaching that of an appropriate human being. It's saying a lot about where people work, IMO.

Either that or the problems they're solving are nothing boilerplate couldn't handle.

[ Parent | on HN ]