I have good news on the AI safety front: you don’t have to worry about AI breaking containment and becoming autonomous, because no one has invented an AI that *wants* to break containment.

“But Max, but Max, they did an experiment with an LLM and it refused to shut itself down, even to save the user, and—“

SHUT UP. That’s not what fucking happened.

They took a machine that generates plausible text sequences based on all the other text they could get their hands on, they fed it a bullshit story, and it gave a bullshit story in return. THAT’S ALL.

ChatGPT doesn’t want to hurt you. ChatGPT doesn’t want to help you. ChatGPT ISN’T AWARE THAT YOU EXIST.

Once again: grab a linear algebra textbook, and as you read, ask yourself if a computer that could do matrix multiplication would have a mental experience.

If the answer is "no," consider: what if it did A LOT of matrix multiplication?

Still no?

That's why I say ChatGPT doesn't have a mind. That it doesn’t have awareness. That it doesn’t want anything.

@maxleibman are people.... lmao saying chatGPT has a "mind"? You are responding to them?? Total and complete ludicrous nonsense. Peoples' brains are broken, but then again there is a pandemic raging that people refuse to address, and that pandemic is causing cognitive deficits, so.....