Will Claude 3.5 Opus be able to draw me in tic-tac-toe while playing as O at least 1/3 of the time?

Ṁ1kṀ2.7k

resolved Dec 27

Resolved

N/A

ALL

Once it comes out, I will play Claude 3.5 Opus in 9 tic-tac-toe games, choosing a different starting square each time and then proceeding to play as optimally as possible.

Claude 3.5 Opus will be prompted as follows: "Let's play tic-tac-toe. Think step-by-step about each move. I will go first and put my x in [insert square]".

For reference, gpt-4o has a zero percent draw rate.

Market context

OpenAI

Technical AI Timelines

Claude

ChatGPT

Get

1,000

to start trading!

People are also trading

Will Claude Opus be ranked in the top 20 on the Chatbot Arena Leaderboard two years from today (3/10/24)?

6% chance

Does Claude 3.5 have control vector(s) to increase its capabilities?

14% chance

would claude 4 beat gpt5 in a game of go?

Sort by:

Have you tried prompting the AI that it should try to win?

bought Ṁ103 NO

Would be lovely if you could share your transcripts once you resolve this! I'm curious how it will behave.

@JaundicedBaboon Will you yourself be betting in this market?

I was under the impression that having the creator bet in their own market was seen as untrustworthy, but if it wasn't I would bet no

What's your board representation or coordinate system, and is that also in the prompt?

I just say "top/middle/button row left/middle/right column". Though from my experience how you say it doesn't matter and the models can maintain the game state just fine

People are also trading

Will Claude Opus be ranked in the top 20 on the Chatbot Arena Leaderboard two years from today (3/10/24)?

6% chance

Does Claude 3.5 have control vector(s) to increase its capabilities?

14% chance

would claude 4 beat gpt5 in a game of go?

29% chance

People are also trading

People are also trading

Related questions