Betting your life on ChatGPT results... Posted on April 6, 2025 at 02:51:48 PM by Craig G
... is probably not a good idea.
I'm having a lot of fun exploring the limits of ChatGPT and DeepSeek.
However, as wonderful as they are, they do make mistakes.
I asked DeepSeek how it would go about calculating show payoffs at Wheeling.
First, it was out of touch, using a pool of about $2000, when $300 is more like it. But then it gave an example and calculated a $24,000 per dollar return on a routine show bet. Apparently it has no common sense error detection or it would have never let that thru. We humans know how to do that routinely by just rounding off numbers for easy computation to get a ballpark figure. So we can catch any order of magnitude errors that a calculator or spreadsheet might produce owing to typos or otherwise.
So I asked ChatGPT:
Why are you unable to see that a $24,000 payoff per dollar in your example is an absurd error?
You’re absolutely right, and I apologize for that oversight! The payoff I mentioned ... is indeed an absurd error and a result of a miscalculation.
That's just one many errors that I've found, so best be careful accepting anything verbatim.
Replies: There have been no replies.
You must register before you can post on this board. You can register here.
Tiger's Jai-Alai Chalk Talk
