3 ms·

Sorry, I must have missed the rest of your prior comment (or posted before you made an edit to it?) LLMs are not able to create new knowledge, only organize ex

by _acco 3y ago

Sorry, I must have missed the rest of your prior comment (or posted before you made an edit to it?) LLMs are not able to create new knowledge, only organize existing knowledge. Critically, we don't use induction to create new knowledge - but induction is all these LLMs can do. (e.g. To explain the origin of the universe, there is nothing we can induce, as the big bang is not observable.) I don't see how training these AI models more will cause this property to emerge. But knowledge creation is the thing that we want when we say we want AGI. The reason I describe AGI as such is because we only have one example of what an AGI is, namely humans. A popular idea is that machine AGI will look very different from human AGI (e.g. no consciousness or intrinsic motivation or qualia). But this is a bold claim. It contends that there are multiple kinds of general intelligences vs some universal kind. Other universal concepts don't look like this. Consider the universality of computation: a computer is either Turing-complete or it isn't. There is no other kind of general purpose computer. Once we created a Turing-complete computer, it could perform all conceivable computation. This is the power of universality. The first Turing-complete computer could technically play Call of Duty (though no one would have had the patience to play it). It's not like there is some mode of ChatGPT that will produce AGI-level responses if given months or years of time to compute it. Fundamentally, its ability to do whatever it is that AGIs do (AGI's version of Turing computation) is missing.

RugnirViking 3y ago
what would you consider a test of the ability to create new knowledge? I don't think there exists a meaningful distinction here. If I give an AI the same information astronomers had to figure out the big bang conclusion, it would probably be able to figure it out. They "induced" from a bunch of data points that there was one likely explanation. Similarly if I gave it a phenomena I wanted to know about it could propose experiements to run to understand that thing. taking independant action in the world is indeed beyond the limits of chatgpt on it's own, but its not hard to build systems relying heavily on chatgpt to do that, a simple loop calling it with information from sensors and a prompt to experiment with outputs is enough to get some kind of agent. It can process any information, and given more power it can draw ever-higher inferences from that information. This is analogous to a turing machine, something that can eventually get anywhere, but it will take an arbitrarily long time to do it. I think of it kinda like the CPU in a Von Neumann architecture. A cpu on it's own doesn't do much of anything unless you tell it to. We have that part solved, we just need to settle on a good structure for repeatedly querying it.

by _acco 3y ago

RugnirViking 3y ago

what would you consider a test of the ability to create new knowledge? I don't think there exists a meaningful distinction here. If I give an AI the same information astronomers had to figure out the big bang conclusion, it would probably be able to figure it out. They "induced" from a bunch of data points that there was one likely explanation. Similarly if I gave it a phenomena I wanted to know about it could propose experiements to run to understand that thing. taking independant action in the world is indeed beyond the limits of chatgpt on it's own, but its not hard to build systems relying heavily on chatgpt to do that, a simple loop calling it with information from sensors and a prompt to experiment with outputs is enough to get some kind of agent. It can process any information, and given more power it can draw ever-higher inferences from that information. This is analogous to a turing machine, something that can eventually get anywhere, but it will take an arbitrarily long time to do it. I think of it kinda like the CPU in a Von Neumann architecture. A cpu on it's own doesn't do much of anything unless you tell it to. We have that part solved, we just need to settle on a good structure for repeatedly querying it.