That’s mostly a philosophical debate. If we create something that is perfectly indistinguishable from actual intelligence, I would call it actual intelligence, but the philosophers might not.
I am perfectly willing to be, but haven’t currently been, convinced that there is a real distinction between natural and artificial intelligence. Until I am convinced though, I’m going to assume there’s no inherent quality about our brain that makes it irreproducible by technology.
ChatGPT emulates intelligence pretty well, but you can still tell the difference between it and a human. So I would say again, we’re not at the point of AGI yet.
But let me ask you, what is the difference between a machine that is perfectly capable of writing intelligent responses to questions and a human writing intelligent responses to questions? (Assuming we’re only measuring intelligence, not things like having a digestive system.) Could you think of a way to tell the two apart, only being able to ask questions to them and receive their responses?
Considering that computers are Turing complete, yes they can, by definition. They can be used to compute anything that can be computed. The question you’re probably really asking is can we make a functional agi with current technology. In a practical sense, no, in a theoretical sense, yes. In practice we can’t because we don’t know how. That knowledge is a form of technology that we haven’t developed yet, though we may have all or most of the pieces available right now. We know that our computers should be able to do it, given enough memory and processing power, but hardware alone doesn’t make an intelligence. You need the software too, and we just don’t know how to make the leap from single purpose tools to general intelligence. Think of it like an airplane. We had all the pieces necessary to make one long before we ever did. We saw birds do it and tried to copy them. We had metal, wood, rope, rubber, cloth, everything you need physically to build a self propelled flying machine, for hundreds or thousands of years, but we didn’t have the underlying principles, a working theory for how to put them together just so. That’s where we are with agi. We have all the raw materials, and some of the complex pieces, but we’re missing things that prevent us from taking that final step into a true agi, however limited.
Can our current models of computing truly create AGI?
There’s no reason to believe they can’t. We’re just not there yet.
As far as I know there is still a big ongoing debate about if there is something fundmental to intelligence that is not just calculations.
That’s mostly a philosophical debate. If we create something that is perfectly indistinguishable from actual intelligence, I would call it actual intelligence, but the philosophers might not.
Well that is still very much a scientific and relevant debate.
Some people will tell you that ChatGPT is intelligent but just because it can write like an intelligent person does not make it intelligent.
I am perfectly willing to be, but haven’t currently been, convinced that there is a real distinction between natural and artificial intelligence. Until I am convinced though, I’m going to assume there’s no inherent quality about our brain that makes it irreproducible by technology.
ChatGPT emulates intelligence pretty well, but you can still tell the difference between it and a human. So I would say again, we’re not at the point of AGI yet.
But let me ask you, what is the difference between a machine that is perfectly capable of writing intelligent responses to questions and a human writing intelligent responses to questions? (Assuming we’re only measuring intelligence, not things like having a digestive system.) Could you think of a way to tell the two apart, only being able to ask questions to them and receive their responses?
Considering that computers are Turing complete, yes they can, by definition. They can be used to compute anything that can be computed. The question you’re probably really asking is can we make a functional agi with current technology. In a practical sense, no, in a theoretical sense, yes. In practice we can’t because we don’t know how. That knowledge is a form of technology that we haven’t developed yet, though we may have all or most of the pieces available right now. We know that our computers should be able to do it, given enough memory and processing power, but hardware alone doesn’t make an intelligence. You need the software too, and we just don’t know how to make the leap from single purpose tools to general intelligence. Think of it like an airplane. We had all the pieces necessary to make one long before we ever did. We saw birds do it and tried to copy them. We had metal, wood, rope, rubber, cloth, everything you need physically to build a self propelled flying machine, for hundreds or thousands of years, but we didn’t have the underlying principles, a working theory for how to put them together just so. That’s where we are with agi. We have all the raw materials, and some of the complex pieces, but we’re missing things that prevent us from taking that final step into a true agi, however limited.
As far as I know there is still a big ongoing debate about if there is something fundmental to intelligence that is not just calculations.