Will Mira's main GPT-4 Sudoku market resolve YES?
119
2.7K
1.2K
resolved Jan 17
Resolved
NO

Note that it can also resolve 50%, in which case this market resolves NO.

See also: /JonathanRay/will-miras-gpt4-sudoku-market-resol

Since the original market should never resolve N/A, only YES/NO/50%, the trio should be fully arbitrageable.

Get Ṁ200 play money

🏅 Top traders

#NameTotal profit
1Ṁ4,926
2Ṁ1,355
3Ṁ1,138
4Ṁ883
5Ṁ824
Sort by:

2 years of hype and the best AI can't solve easy sudoku puzzles for all that money put into it

bought Ṁ1 YES at 13%
predicted YES

@Seeker Well, some

bought Ṁ1 of YES

@EmilyThomas what is the "turns" column?

predicted YES

@Seeker The number of chat responses from GPT-4 needed to solve the puzzle (or to make a mistake).

predicted YES

@Mira im slightly impressed then. It failed after 4 in January 3 but went to 27 turns on January 2 without mistake. I'd probably fail if I didnt know how to play sudoku before 20 turns

bought Ṁ17 of NO

thank you