I meant signal manipulation, but sure, that includes manipulating humans since we are just another signal input. That isn’t a statement on values, just an objective description of how it works.
That “intelligence” will tell you how it is not as intelligent if you ask directly. This is not because it’s intelligent and KNOWS it’s limitations but because it picked up on my signal for what I wanted and gave something resembling that to me. You could ask it the opposite thing and arrive at a very different conclusion.
https://claude.ai/share/97859aae-beb2-4245-a8b1-460122ee4aa1
I would even say the analogies it gave aren’t entirely accurate. A person reading a book experiences it very differently than an LLM. And this is a key insight to an element of the Turing test where you ask it to explain some poetry (not simply write a poem).
You can lower the bar for what you consider intelligent, but that doesn’t make the LLM any more capable.
And no, I am not talking about emotions in regards to reading a book. Do you simply experience a book as a set of words and emotions attached to those words? I think you are selling yourself short in order to put LLMs on a pedestal.
Because signal matching and manipulation is not intelligence unless you’re willing to admit that is all you are capable of.
AI tools don’t actively manipulate people. They’re just confident about their answers as they should be.
I think they meant signal matching and signal manipulation
I meant signal manipulation, but sure, that includes manipulating humans since we are just another signal input. That isn’t a statement on values, just an objective description of how it works.
I still don’t understand how it cannot be as intelligent as us.
That “intelligence” will tell you how it is not as intelligent if you ask directly. This is not because it’s intelligent and KNOWS it’s limitations but because it picked up on my signal for what I wanted and gave something resembling that to me. You could ask it the opposite thing and arrive at a very different conclusion. https://claude.ai/share/97859aae-beb2-4245-a8b1-460122ee4aa1
I would even say the analogies it gave aren’t entirely accurate. A person reading a book experiences it very differently than an LLM. And this is a key insight to an element of the Turing test where you ask it to explain some poetry (not simply write a poem).
While I agree that it doesn’t have a physical experience, I don’t think a physical experience is necessary to speak to someone.
I meant LLMs and people experience reading a book differently in the internal sense. Or at least for some people.
Of course! We hallucinate being in the character’s place. The LLM can’t experience that because it has no emotions.
In any case, unless you’re talking about emotional intelligence, I don’t think that’s necessary to be intelligent.
You can lower the bar for what you consider intelligent, but that doesn’t make the LLM any more capable.
And no, I am not talking about emotions in regards to reading a book. Do you simply experience a book as a set of words and emotions attached to those words? I think you are selling yourself short in order to put LLMs on a pedestal.