Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

“Don Knuth Plays with ChatGPT” but with ChatGPT-4

gist.github.com

Post reply on HN
Print view View original

31–40 of 141 posts

Re: “Don Knuth Plays with ChatGPT” but with ChatGPT-4

#31

Post by beanaroo »

hannasm wrote: post #19
bee_rider wrote: post #10

Earlier quoted context omitted.

That is a weird way of spelling mergesort.

I believe radix sort belongs first in this list.

The most elegant is certainly sleepsort. Maybe not the most efficient, but definitely elegant.

Re: “Don Knuth Plays with ChatGPT” but with ChatGPT-4

#32

Post by bombcar »

mod50ack wrote: post #24
axpy906 wrote: post #17

Nailed every one. Some by saying not possible to answer but still.

Didn't nail the Rodgers and Hammerstein one; it still doesn't understand the reference to the ballet or that the "themes" in the question are musical.

I wouldn’t be surprised if half the Internet does not know that a ballet is part of a larger show.

Re: “Don Knuth Plays with ChatGPT” but with ChatGPT-4

#33

Post by gfodor »

sebzim4500 wrote: post #25
axpy906 wrote: post #17

Nailed every one. Some by saying not possible to answer but still.

Got the 'five character word' question wrong. Admittedly I also thought it was correct at first glance but then went back when someone called it out in another comment.

Language models struggle specifically with token games like this, since they can’t see them at that resolution or something.

Re: “Don Knuth Plays with ChatGPT” but with ChatGPT-4

#34

Post by bee_rider »

hannasm wrote: post #19
bee_rider wrote: post #10

Earlier quoted context omitted.

That is a weird way of spelling mergesort.

I believe radix sort belongs first in this list.

Performance-wise, maybe, but mergesort is clearly the most elegant/beautiful sorting algorithm. Nothing tricky going on, just a couple sorted lists being merged. Plus everyone loves a stable sort.

Re: “Don Knuth Plays with ChatGPT” but with ChatGPT-4

#35

Post by iudqnolq »

ec109685 wrote: post #26
iudqnolq wrote: post #12

Earlier quoted context omitted.

It also fails to write a sentence with only five character words.

It did get closer. For that type of query you can ask it check its work and can usually triangulate on correct answer within a single prompt, eventually.

I would be cautious of a Clever Hans effect there. If you repeat the question until you get the right answer you're providing the AI with significant extra information.

Re: “Don Knuth Plays with ChatGPT” but with ChatGPT-4

#36

Post by nearbuy »

iudqnolq wrote: post #12
ec109685 wrote: post #6

Interesting both completely whiff on the number of chapters in the Haj.

It also fails to write a sentence with only five character words.

Still fairly impressive. Probably better than most people could do if given 60 seconds, but probably worse than most people if given 10 minutes.

Re: “Don Knuth Plays with ChatGPT” but with ChatGPT-4

#38

Post by paulddraper »

nearbuy wrote: post #36
iudqnolq wrote: post #12

Earlier quoted context omitted.

It also fails to write a sentence with only five character words.

Still fairly impressive. Probably better than most people could do if given 60 seconds, but probably worse than most people if given 10 minutes.

I don't think that is true.

Re: “Don Knuth Plays with ChatGPT” but with ChatGPT-4

#39

Post by ResearchCode »

ryanseys wrote: post #7

It now knows to communicate that the NASDAQ doesn't operate on Saturdays.

Did it know that before the last LLM failure was posted on Twitter or Hackernews? Trawling tech media for LLM failures can be assumed to be part of the "human feedback".
Post reply on HN