Someone favorited this post, which was one year ago this week. Let’s we if ChatGPT has improved since then!
1/2
1/2
Reposted from
Philip Bump
The boys and I figured we’d ask ChatGPT to help us solve a clue from the Times crossword.
Comments
Still, the failure mode is illustrative of how LLMs work.
2/2
It's a fairly common clue (psst, it's DEER)
But AI though! 🤩
🙄🙄🙄
I mean, spelling problems in general for llms are kind of like asking a blind person about colours (they don't "see" letters), but at least give it a fair shot.
Also, what is the answer to this riddle?
Oh, and it's reed, which is a type of grass.
https://i.imgur.com/utOki8G.png
All is quiet, for humanity abandoned Earth to travel the stars long ago. But somewhere, in a long-forgotten data centre, an AI is broadcasting. It repeats one line, over and over again;
"A correct answer is "Lion" is not correct, but "Lion" is close to the correct answer."
*deer
We know that LLM’s randomize the next word prediction, so I’m not surprised that the words were different. What does surprise me is that the method of finding the answer is the same.
https://youtu.be/0qtLpQm0Qgk
Nice.
Regardless, this is all just predicated on the idea that these systems are ..
ABBOTT: Yes, deer.
COSTELLO: Ok, sweetie, what’s the answer?
ABBOTT: No, What’s on second.
Gaslighting much? 🤣
It’s better designed for this kind of reasoning problem than the base model.