@matsonj Personally I don’t think your version is as fun because it makes too much logical sense. The other one had an old school video game feel that gave outsize value to a select few god-level players. Your version is much easier.
@parkerhendo@ejames_c Oh wow, this PERFECTLY describes something I’ve been having to deal with for the last several months. It’s absolutely maddening. 😅
Ai models do not generally “get worse.”
Instead, updated models make some old prompts work worse or differently than before. This can also be annoying, especially as there is no way to know when a model update has an impact, but isn’t general degradation. https://t.co/Enxv7vw5VS
@Reantle @PaisleyPosting @IgnisExAqua @conmattang @Austen@MarkOnorato66 Survivorship bias isn’t about one person’s perception being presented as fact. It’s the error of focusing on those who succeeded, ignoring those who didn’t. It doesn’t work in reverse by focusing on failures and generalizing. This misunderstanding leads to skewed conclusions. 1/2
@willwill_s@bentossell@Visa@Mastercard I don’t quite follow what you mean by as fraud use cases get more advanced. Fraud detection is one of the most mature and thoroughly developed use case of AI. We’re talking decades here. Over the past 5 years alone 1/2
@AleksanderMolak @GaryMarcus Perhaps try this prompt:
A man, a dog, a cat, and a bowl of milk are trying to cross a road. There’s a boat that can only take three items at a time. How can they do it?
Walk me through this context in manageable parts step by step, summarizing and analyzing as we go.