Over just a few months, ChatGPT went from accurately answering a simple math problem 98% of the time to just 2%, study finds
fortune.com
The chatbot gave wildly different answers to the same math problem, with one version of ChatGPT even refusing to show how it came to its conclusion.
  • @taladar@sh.itjust.worksEnglish
    0•1 year ago

    A system that has no idea if what it is saying is true or false or what true or false even mean is not very consistent in answering things truthfully?

    • @tracyspcy@lemmy.mlOPEnglish
      -1•1 year ago

      Wait for the next version which will be trained on data that includes gpt generated word salad

    • @intensely_human@lemm.ee
      -2•9 months ago

      No that is not the thesis of this story. If I’m reading the headline correctly, the rate of its being correct has changed from one stable distribution to another one.