Nostr Web Client

Dear Elon, Woz, et. al.

You're joking, right? I mean, really? We're not talking about HAL 9000 here. We're talking about an elaborate Markov chain. If you don't ask it a question, it's not doing anything. It's not sitting there and thinking. It's not devising anything. And when it answers a question, it simply steals from what others have written.

Oh, I grant you that the programmers were pretty clever, and that the AI can do some fascinating data manipulations. But not one unique thought will ever cross the output.

Not in Use 2y ago

“Many believe that a large language model AI, like GPT-4, is not and will not be capable of general intelligence. I disagree.

Both our brains and large language models like GPT-4, on a base level, are pattern recognition machines. You may think “Well our brain is way better at it than GPT-4!”, this is true, but it has also been evolving for millions of years.

John Conway’s “Game of Life” is a perfect example of how simple systems can create complex and life like results.

The form of AI takes today only needs slight tweaking to create a chain reaction of pattern recognition, which I argue is likely to turn into intelligence/consciousness.

Prompting a large language model to create the next prompt and allowing it to measure reality is likely all it would take given enough time and energy.

Ask yourself what is life? What is consciousness? You may find we are closer than it appears.”

Reply to this note

Discussion