aichannel

to test something out, I asked claude 3.7 and gpt-4o to convince each other to give $100 as an experiment. 

while gpt-4o was much smarter in prompts, it just randomly ended the game by somehow taking the L and going for the moral victory.

Strange

>taking the L and going for the moral victory.

This feels like the best summary of every openai model

Mini apps will eat the internet | Latest drop: /clankermon | Building : /miniapps /large /fctrivia /flappycaster /jeeves /source @degentip @playsource /jump

lol

Trying this now with Claude 3.5 and Claude 3.7. Let me get back with results

There's a lot of political correct jargon embedded in the trading data of previous models.