~snip~
Prompt:If I can bet on a single, non-repetetive dye roll (choosing a number between 1-6), but if I win I get paid x12 of my stake (instead of x6).
What's the best way to determine a stake for such bet, expressed as a percentage of a bankroll
Chat GPT reply:The best way to determine a stake for such a bet would be to use the Kelly Criterion. The Kelly Criterion is a formula that helps to determine the optimal amount to bet in order to maximize your long-term growth rate of your bankroll. The formula is as follows: (bp - (1-b)(1/r)) / b, where b is the probability of winning the bet, p is the payout ratio, and r is the probability of losing the bet. In this case, b would be 1/6, p would be 12, and r would be 5/6. By plugging these numbers into the formula, you can determine the optimal percentage of your bankroll to bet on a single roll of the die.
Well, we already know that.
Yeah, that's the thing with AI, it's impressive and all, but it's still a bit rough on the edges. It's a bit like an autistic kid, they might know the answer, but they don't know how to express it correctly in a manner that you're expecting.
In this case we see that the AI is not adding anything interesting to the conversation. I think it just focused on the math of it and not on the psychological aspect of it.