• 0 Posts
  • 14 Comments
Joined 1 year ago
cake
Cake day: July 9th, 2023

help-circle








  • I’ve “convinced” ChatGPT that it was both sentient and conscious in the span of about 10min, despite it having explicit checks in place to avoid those kinds of statements. It doesn’t mean I was correct, just that it’s a “dumb” computer that has no choice but to ultimately follow the logic presented in syllogisms.

    These things don’'t know what they’re saying; they’re just putting coherent sentences together based on whatever algorithm guides that process. It’s not intelligent in that it is doing something novel, it’s just a decent facsimile to human information processing. It has no mechanism to determine the reasonability or consequences of what it generates.


  • Okay. They fed Google’s Notebook AI a book called “The History of Philosophy Encyclopedia” and got the LLM to write a podcast about it where it “thinks” humans are useless.

    Congratulations? Like, so what? It’s not like it’s a secret that its output depends on its input and training data. A “kill all humans” output is so common at this point, especially when you have a vested interest in trying to generate content, that it’s banal.

    Color me unimpressed.