Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Could o1 have helped the hunters finally reach it?


My experience with using LLMs for things like MIT Mystery Hunt is touch and go.

On most of them I've tried it doesn't seem to do much, but I do now use them to try and get crossword clues where I know bits that are often too abstract for crossword solvers.

e.g. "a word that is six letters, is related to royalty, and has a state abbreviation in it" (this isn't a real clue, just an example of a clue that an LLM is much better suited than something like Nutrimatic or a crossword solver)

I would be curious to hear if / how others us LLMs for abstract riddles/puzzles like that though.


I tried, it solved a few steps and was quite impressive! It's really good at riddles and associating random concepts.

It got stuck when it had to calculate directions though.


Not sure why you got downvoted.

It is not crazy to imagine that LLMs could have helped explore the solution space.

(not necessarily solving it directly)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: