You are viewing a single comment's thread from:

RE: The Ghost in the Machine: The Prophecies of AI

in Reflectionslast year (edited)

God, all knowing and all powerful, had only one thing left he didn't know: What was it like to not exist.

Richard Feynman said you can't understand things unless you can build them that's the real test of do you understand it. And AI can understand what humans can't. Even humans can't sense such as 128-dimensional world, we are stuck in 3-d. It's already inevitable AI will manipulate humans sooner in time.

Furthermore, as an AI master's student, I believe being a human to measure existence is either 1 or 0. We exist so there is non-exist exist. There is an experienced form(computed) and not experienced(uncomputed) states. This is an information theory called binary, which is all about computation thus AI.

Sort:  

AI can understand what humans can't.

Well... AI doesn't really 'understand' anything. That's its primary flaw. It's just pulling up data. But it has no concept of which it speaks.

AI defeated the greatest champions at the board game GO. But researchers discovered they can take advantage of the fact that it literally has no concept of a board, pieces or the rules of the game. Ultimately, they taught amateur players to defeat this top super AI machine 95% of the time with some exploitative tricks.

The same applies across the board. It's useful and powerful, but I wouldn't want to conflate that with 'understanding'. At least for some years yet. So far, the only way to get around these issues has simply been to feed it more data, which is just more of the same. There's no barrier being crossed into sentience... so far

Well... AI doesn't really 'understand' anything. That's its primary flaw. It's just pulling up data. But it has no concept of which it speaks.

In a lecture, we tested whether chatgpt can deal with sarcastics or indirect way of saying, that was like Bob and Mary watched a movie in cinema. After watching, Bob said "I think the movie was great, waht do you think?" Mary said " I think popcorn was tasty.". And the lecturer asked to chatgpt 'why Mary is talking about popcorn when Bob is talking about the movie?' Then chatgpt said 'Mary may try to be polite in replying she was not enjoying the movie." At that moment the lecture said "we're doomed."

Ultimately, they taught amateur players to defeat this top super AI machine 95% of the time with some exploitative tricks.

I doubt they defeated AlphaZero now it is at a stage with odds of winning 100% over human players. It generates the map a few hundred steps ahead every time the opponent turned his time.