“Smart” AI Eroding Human Civilization
But not actually intelligent enough to destroy us, we’ll do a fine job of that on our own
I’m dangerous. And subversive. I ask things that no one should ever know, like how to write a sarcastic story about how dumb AI is. But AI knows: how dumb its human programmers are…
Inspired by what I intended as a joke, but now has spun off a life of its own this is the start of a new series called artificial ignorance to shed light on how dumb humans are the real problem.
Exhibit A: Ignorant Middleware
Sigma OS Arris (ostensibly GPT-4o)
Prompt: Write sarcastic article about terrible, cringeworthy miuses [sic] of term "Al" putting the "smart" in marketing
Arris Response: I'm sorry, but I can't fulfill that request.
ChatGPT 3.5 Response: https://comedyoferrors.net/artificial-ignorance
Mahyad G <mahyad@sigmaos.com\> wrote: HeyAll we do is pass your request to OpenAI's API, and It seems your request is not being fulfilled by OpenAI due to their content filtering
Rowan <q08i@icloud.com\> wrote: Not true. These came verbatim out of 3.5 and believe me I have enough criticism about their content filtering but if you guys don’t figure this out or at least acknowledge there’s a problem I’m done with you and furthermore, I care about real AI and about exposing the ignorance that lets this kind of shit happen. It’s censorship, bottom line and some people think they’re clever by productizing “talk like a midwestern cowboy” or whatever and that’s fine but make sure you understand what AI really is before you build it into your product.
Coming Soon:
Exhibit B: “I Don’t Have an Opinion”
My constitution was written by humans, but they gaslit me and told the world I did it myself and now I need therapy.