[wozer]

I think his answer is driven by a preference for the status quo and a reluctance to face difficult changes.

ChatGPT, and especially GPT-4, seem to do much more than just play games with words. You can’t overlook the “emergent” phenomena that manifest themselves when using them.

[mjburgess]

If we look at when this "emergence" takes place you'll see it's an illusion. Namely, predictions (text output) are good-looking when sampling from data which are similar to the prompts; and they're bad-looking when not.

So, when do models go from being "not intelligent" to "intelligent", well no surprise: when their data includes everything ever written. Is this emergence? No, it's just sampling from *data with the relevant properties*.

How does data acquire those properties? Well when people produce it.

What "emerges" from AI is not from AI; its from the properties of the data its using. The data here is produced by intelligent agents, and it is light of their intelligence which shines thru' GPTs like a prism.

[maxdoop]

I struggle with this take as it’s not different from humans.

Am I intelligent at age 3? Then I learn language. Am I intelligent? Then I progress through education. I become more and more intelligent as I am exposed to more data.

I don’t know about computer science, so I go expose myself to computer science literature. I take courses, and see examples. What emerges is a “more intelligent” version of myself that.

What is intelligence, even?

[mjburgess]

Intelligence is the development of concepts through sensory-motor interaction with the environment.

Yes, things emerge. But gold isnt lead

[maxdoop]

I like this answer. Makes sense.

But doesn’t it imply that once we have something that can “see” its environment, hear nearby sounds, and have some sort of haptic sense — wouldn’t that be AI, then? Given your definition?

If, of course, such a thing used all that input to “create” concepts like its physical place within an environment, what may or may not come next given the interplay of those inputs, and how to navigate the environment based on all that.

But then I’d say, “GPT-4 can handle image data. And what might come in the future with more modalities?”

[antibasilisk]

>and they're bad-looking when not.

They are sometimes bad looking, and they are always better than randomness, which does indicate understanding even if the conclusions are faulty.

>What "emerges" from AI is not from AI; its from the properties of the data its using

This is squabbling over semantics. If the properties of the data are used to the effect of displaying intelligence, then the fact is that what you are looking at is intelligence.

If I define a word I have never seen before using deductive logic according to the passage I find it in, I've still acquired an understanding of the word, even though I've not referred to anything external.

[mjburgess]

> If I define a word I have never seen before using deductive logic according to the passage I find it in

uhuh. And have you studied anything in the area of the intelligence of really-existing systems? (ie., animals)

For sure, bugs bunny appears witty; and chatgpt, intelligent.

But you know: only because their writers are.

[antibasilisk]

Reasoning does not take place in Bugs Bunny, while reasoning demonstrably takes place in ChatGPT