The Level1 Show Podcast: 2/12/2024
Added 2024-02-12 21:31:55 +0000 UTCComments
Gpt 3.5 got it wrong but gpt 4 got it right every time I tried.
Jacob Wood
2024-02-18 19:29:52 +0000 UTCThe rock paper scissors test is nice. Another one is: "I have 5 oranges today. Last week I ate 3. How many oranges do I have now?" Almost any chatbot fails at it. If they succeed, open a new chat and try again, they fail, none is consistent. Tested with gpt 3.5, 4, bing copilot, genimi bard. They instead teach you in basic math, and cheer your healthy diet. Also, Bard costs 22+ USD per month, 2TB storage alone costs less than 10. But I guess people will fail at that math because of previously taught example.
Staatlicher Nagetierfunktionär
2024-02-13 10:38:13 +0000 UTC
