Re:Don Knuth Plays with ChatGPT
by furyofantares on Saturday May 20, 2023 @01:18PM
(#36013722)
I'm aware, they still posted an example of something that's already gone away. Many of Knuth's examples are also very noticeably better with GPT-4 and it's a shame it wasn't used, because some of the examples aren't.
Their post without the example stands well on its own IMO and is hampered by what is non-sequiter at best and misinformation at worse tacked onto the end.
the example given by the parent isn’t even an issue with GPT-3 or 3.5, never mind 4. it’s just a quirk of chatGPT. if you access any of the models via the API, you wouldn’t get this problem at all, because you control the system prompt
Re:Don Knuth Plays with ChatGPT
by CamperBob2 on Saturday May 20, 2023 @02:19PM
(#36014288)
It's still there in GPT-4.
There's less of it, though.
And only the first couple of time derivatives matter.
is it?
I can tell you for a fact that if I put "don’t use words like please, sorry, apologize, etc. speak plainly" in the gpt-4 (or 3 or 3.5) system prompt, it will not use those words in its completions
Re:Don Knuth Plays with ChatGPT
by CamperBob2 on Saturday May 20, 2023 @04:51PM
(#36015494)
I'm confused
I would in general agree with the parent that the, as I like to call it, "convincing bullshit" will not go away. It's still there in GPT-4.