3 ms·

> What I'm saying is that HOW it thinks might already be quite close in essence to how WE think. How would one be able to prove this? Nobody knows how we think

> What I'm saying is that HOW it thinks might already be quite close in essence to how WE think. How would one be able to prove this? Nobody knows how we think, yet. All one can say is that what GPT-4 outputs could plausible fool another human into believing another human wrote it. But that's exactly what it's designed to do, so what's interesting about that? > If you concede that our brain is "simulatable", It could be. Maybe. It might be that's what the universe is doing right now. Does it matter? We're talking about writing an emulator on a Harvard-architecture computer that can fully simulate the physics and biological processes the make up a human brain. By interpreting this system in our emulator we'd be able to witness a new human being that is indistinguishable from one that isn't simulated, right? That's not what GPT is doing. Not even close. It turns out there's more to being human than being a register VM. Ever get punched in the face? Bleed? Fall in love? Look back on your life and decide you want to change? Write a book but never show it to anyone? Raise a child? Wonder why you dreamt about airplanes on Mars with your childhood imaginary friend? Why you hate bananas but like banana bread? Why you lie to everyone around you about how you really feel and are offended when others don't tell you the truth? It's not so simple.

myrmidon 4y ago
> We're talking about writing an emulator on a Harvard-architecture computer that can fully simulate the physics and biological processes the make up a human brain. By interpreting this system in our emulator we'd be able to witness a new human being that is indistinguishable from one that isn't simulated, right? My point is: if you don't believe that there is magic pixy dust in our brains, then this would NECESSARILY be possible. It would almost certainly be HIGHLY inefficient-- the "right way" to do AGI would be to find out which algorithmic structures are necessary for human level "performance", and implement them in a way that is suitable for your VM. I'm arguing that GPT4 is essentially the second approach-- it lacks features for full human level performance BY DESIGN (e.g. requires pre-training, no online learning, etc.), but there is no reason to assume that the way it operates is fundamentally different from how *parts* of OUR mind work. > It turns out there's more to being human than being a register VM. Ever get punched in the face? Bleed? Fall in love? Look back on your life and decide you want to change? Write a book but never show it to anyone? Raise a child? Wonder why you dreamt about airplanes on Mars with your childhood imaginary friend? Why you hate bananas but like banana bread? Why you lie to everyone around you about how you really feel and are offended when others don't tell you the truth? I don not understand what you are getting at here. I consider myself a biological machine-- none of this is inconsitent with my worldview. I believe that a silicon based machine could emulate all of this if wired up properly. PS: I often talk with people that explicitly DONT believe into the "pixy dust in our brains" (call it soul if you want), but on the other hand they strongly doubt the feasibility of AGI-- this is internally inconsistent and simply not a defensible point of view IMO.
- agentultra 4y ago
  > I'm arguing that GPT4 is essentially the second approach Ok, so then it is an algorithm that simulates a specific behaviour that produces plausibly human-level results. My point is that this is not thinking, smart, or "general intelligence." Let's say I write an algorithm that can also produce text. It's not an implementation of the specification for GPT-4 but something novel. It takes the exact same inputs and produces outputs that I share with you and claim is produced by GPT-4. And lo, success, you can't tell if it was produced by GTP-4 or my algorithm. You claim it's the same thing as having GPT-4, right? If you can't tell the difference it must be the same thing. Big deal. We can write computer programs that perform better than humans at chess, go, and now can write more text than us. We knew this was possible before we even begun on this endeavour. It's still not intelligent, conscious, smart, or anything resembling a complete human. It's merely an algorithm that does one specific task. > I don not understand what you are getting at here. I've proven my point then. There's more to the human experience than what can be simulated on a silicone chip and it doesn't have to do with hand-waving away all the complexity of reality as "magical pixie dust." Take physical trauma. The experience of which by one human is not merely a fact. It is felt, it is reflected upon, and it is shared in the DNA of the person that experience it with their descendants. We have science investigating how trauma is shared through generations and the effects it has on our development. You are more than a machine with inputs and outputs.
  - myrmidon 4y ago
    > My point is that this is not thinking, smart, or "general intelligence." Why not? I would already, without hesitation, describe GPT4 as strictly more intelligent than my cat and also all gradeschoolers I've ever known... Maybe some adults, too- depends on your exact definition of intelligence. > Let's say I write an algorithm [...], you can't tell if [input] was produced by GTP-4 or my algorithm. Sure, I'd call your algorithm just as clever as GPT4 and approaching adult human levels of intelligence. > It's still not intelligent, conscious, smart Why not? What do these mean to you?
    - agentultra 4y ago
robwwilliams 4y ago
I think you would love to read Mark Rowlands’ The Philosopher and the Wolf. He asks these questions and like all if us struggles with answers. https://www.goodreads.com/book/show/8651250 https://www.goodreads.com/book/show/8651250
krickkrack 4y ago
"Nobody knows how we think, yet." Then how can you confidently say we don't think 'like' Transformers/Attention/Statistical models/etc/etc?

myrmidon 4y ago

> We're talking about writing an emulator on a Harvard-architecture computer that can fully simulate the physics and biological processes the make up a human brain. By interpreting this system in our emulator we'd be able to witness a new human being that is indistinguishable from one that isn't simulated, right? My point is: if you don't believe that there is magic pixy dust in our brains, then this would NECESSARILY be possible. It would almost certainly be HIGHLY inefficient-- the "right way" to do AGI would be to find out which algorithmic structures are necessary for human level "performance", and implement them in a way that is suitable for your VM. I'm arguing that GPT4 is essentially the second approach-- it lacks features for full human level performance BY DESIGN (e.g. requires pre-training, no online learning, etc.), but there is no reason to assume that the way it operates is fundamentally different from how *parts* of OUR mind work. > It turns out there's more to being human than being a register VM. Ever get punched in the face? Bleed? Fall in love? Look back on your life and decide you want to change? Write a book but never show it to anyone? Raise a child? Wonder why you dreamt about airplanes on Mars with your childhood imaginary friend? Why you hate bananas but like banana bread? Why you lie to everyone around you about how you really feel and are offended when others don't tell you the truth? I don not understand what you are getting at here. I consider myself a biological machine-- none of this is inconsitent with my worldview. I believe that a silicon based machine could emulate all of this if wired up properly. PS: I often talk with people that explicitly DONT believe into the "pixy dust in our brains" (call it soul if you want), but on the other hand they strongly doubt the feasibility of AGI-- this is internally inconsistent and simply not a defensible point of view IMO.

agentultra 4y ago
> I'm arguing that GPT4 is essentially the second approach Ok, so then it is an algorithm that simulates a specific behaviour that produces plausibly human-level results. My point is that this is not thinking, smart, or "general intelligence." Let's say I write an algorithm that can also produce text. It's not an implementation of the specification for GPT-4 but something novel. It takes the exact same inputs and produces outputs that I share with you and claim is produced by GPT-4. And lo, success, you can't tell if it was produced by GTP-4 or my algorithm. You claim it's the same thing as having GPT-4, right? If you can't tell the difference it must be the same thing. Big deal. We can write computer programs that perform better than humans at chess, go, and now can write more text than us. We knew this was possible before we even begun on this endeavour. It's still not intelligent, conscious, smart, or anything resembling a complete human. It's merely an algorithm that does one specific task. > I don not understand what you are getting at here. I've proven my point then. There's more to the human experience than what can be simulated on a silicone chip and it doesn't have to do with hand-waving away all the complexity of reality as "magical pixie dust." Take physical trauma. The experience of which by one human is not merely a fact. It is felt, it is reflected upon, and it is shared in the DNA of the person that experience it with their descendants. We have science investigating how trauma is shared through generations and the effects it has on our development. You are more than a machine with inputs and outputs.
- myrmidon 4y ago
  > My point is that this is not thinking, smart, or "general intelligence." Why not? I would already, without hesitation, describe GPT4 as strictly more intelligent than my cat and also all gradeschoolers I've ever known... Maybe some adults, too- depends on your exact definition of intelligence. > Let's say I write an algorithm [...], you can't tell if [input] was produced by GTP-4 or my algorithm. Sure, I'd call your algorithm just as clever as GPT4 and approaching adult human levels of intelligence. > It's still not intelligent, conscious, smart Why not? What do these mean to you?
  - agentultra 4y ago

robwwilliams 4y ago

I think you would love to read Mark Rowlands’ The Philosopher and the Wolf. He asks these questions and like all if us struggles with answers. https://www.goodreads.com/book/show/8651250 https://www.goodreads.com/book/show/8651250

krickkrack 4y ago

"Nobody knows how we think, yet." Then how can you confidently say we don't think 'like' Transformers/Attention/Statistical models/etc/etc?