4 ms·

Why is it not an AGI? I keep seeing naysayers complain that it is not, but nobody actually articulates exactly what makes it a non-AGI in a manner that is suff

Why is it not an AGI? I keep seeing naysayers complain that it is not, but nobody actually articulates exactly what makes it a non-AGI in a manner that is sufficiently robust to argumentation. I am willing to play the GPT's advocate and have this argument here provided that I do not get rate limited. Edit: I'd like to respond to some of the comments but I have been rate limited :')

catoc 4y ago
Not trying to be a naysayer or downplay how cool GPT4 is, but the term 'AGI' has existed for decades and is suffering from general confusion and inflation (again like 'crypto' now being read as 'coins' or even 'NFT' by the general public). See eg: https://en.wikipedia.org/wiki/Artificial_general_intelligence https://en.wikipedia.org/wiki/Artificial_general_intelligenc... or better read Bostrom's book "Super Intelligence"
wizofaus 4y ago
I've posted a number of times why I don't think current GPT tech classifies as AGI and certainly not human-like intelligence, as have others. Ask it to solve novel mathematical and logic problems, for a start (even ones you'd expect high school kids to manage), or simple word puzzles (anagrams and the like - again, avoiding "known" examples that it's likely to have come across in its training data). Further, an AGI should logically attempt to "improve" its understanding of the world during a conversation, yet chat GPT never initiates a topic or asks you anything at all (even to clarify what you mean). It also repeatedly reminds you that it is only a language model.
- antibasilisk 4y ago
  Machines can never obtain human like intelligence because they are foundationally not human like, that doesn't mean they lack intelligence. That's like saying dolphins lack intelligence.
  - wizofaus 4y ago
    I wouldn't claim ChatGPT lacks intelligence though. Clearly in terms of language comprehension and generation it does so.
  - z3c0 4y ago
    False equivalence. Dolphins have exhibited general problem-solving skills in a myriad of situations, and thus qualify as a general intelligence.
    - wizofaus 4y ago
      Indeed, I had started to write something similar. And pretty clearly animals like Dolphins or Chimps when they exercise their general intelligence don't do it by manipulation of language tokens. BTW I tried to get Chat-GPT to solve a problem chimpanzees have been observed being able to solve (using an inverted bowl to stand on in order to reach some bananas): "I'm sorry, but it is not biologically plausible for a chimpanzee to use a bowl of water to access bananas on a tree"
      - dave84 4y ago
        GPT-4 had no issues answering this.
        wizofaus 4y ago
        Presumably because it's similar enough to descriptions of problems and how they were solved that are in its training data.
    - antibasilisk 4y ago
      As has PaLM-E
  - pixl97 4y ago
    I believe you're correct, but for possibly the wrong reasons. I believe that AI will be able to simulate a lot (if not most) humanlike action in the near future. The particular problem here is they will also simulate far more non-humanlike intelligence making the alignment issue an even larger problem than it already is.
- ekleraki 4y ago
  > Ask it to solve novel mathematical and logic problems How exactly do you define novel tasks and problems? My brother has had it do his programming homework (he's a teenager), written in Greek mind you and therefore novel task to the first-degree, and it succeeded. Thus it solved a logic task [1]. I have had ChatGPT parse math I had written in LaTeX and reach a correct result, do modifications and so on. > Further, an AGI should logically attempt to "improve" its understanding of the world during a conversation, yet chat GPT never initiates a topic or asks you anything at all (even to clarify what you mean). It also repeatedly reminds you that it is only a language model. Anecdata: I asked Bing (GPT4) to answer a few stuff and it asked for clarification twice in a row. The fact that it repeatedly clarifies its prompt is not a limitation of the underlying model as much as it is a limitation of the interface that injects a prompt. Dan and all the other jailbreaks out there bypass that. [1] https://en.wikipedia.org/wiki/Curry%E2%80%93Howard_correspondence https://en.wikipedia.org/wiki/Curry%E2%80%93Howard_correspon...
  - UncleEntity 4y ago
    > I have had ChatGPT parse math I had written in LaTeX and reach a correct result, do modifications and so on. Interesting, didn’t know it could do that. Wondering if it could also follow along with the lemmas in various papers and reach reasonable conclusions.
    - PartiallyTyped 4y ago
      It depends, I have used Copilot to modify LaTeX and write segments of my thesis, and it works surprisingly well, though its arguments are not very precise.
  - wizofaus 4y ago
    https://medium.com/illumination/i-put-chatgpt-to-the-test-heres-what-it-couldn-t-handle-5897de20481d https://medium.com/illumination/i-put-chatgpt-to-the-test-he... has some examples of simple maths/logic problems it can't really solve. It'd be good to know how much better chatgpt-4 is at them.
Karunamon 4y ago
One major reason it doesn't qualify is that it is impossible for it to learn from its input. It's right there in the name: generative pre-trained transformer GPT basically has two major modes of operation right now, training, which is done by openai, and output, which is what you are doing with the API and with ChatGPT. The model does not update its "brain", its parameters and weights, based on the conversation you have with it. Even the "conversation" presented by their chat UI is a little misleading, what's happening is that the entire conversation is being replayed into the model every time you hit send. There is also a hidden input provided by openai, responsible for the "I am a large language model" stuff, anything said early in the conversation can constrain the later output, sometimes to the point of uselessness.
- PartiallyTyped 4y ago
  > One major reason it doesn't qualify is that it is impossible for it to learn from its input. It's right there in the name: generative pre-trained transformer Both chatGPT and GPT models exhibit in-context learning, just as you do. The difference is that during your context updates modifications to your synapses occur as a consequence of your brain's activations. Let's entertain a hypothetical scenario where I am a 5D creature, and you are (3 Space + 1 Time)D human. I could go back to any point in your time and have any conversation that I wish with you. Each time I do that, I am effectively resetting your brain's state to that particular point in time. The analog here is that for ChatGPT and the likes, time does not exist in the manner that it does for us, they are purely abstract in that they don't even need physical form to exist, it's a collection of ones and zeros that we store in such form because it is convenient. You can very well store the parameters of the GPT models in any other medium capable of holding information and actually execute the model. You can even compute gradient updates, or hook the models up to external data sources. In the same vein, you "could" do the same for any human brain no matter how computationally prohibitive it may be. You'd still classify the human brain as a general intelligence. Furthermore, we classify humans with long-term amnesia or inability to form new memories generally intelligent even though they only exhibit in-context learning.
  - Karunamon 4y ago

catoc 4y ago

Not trying to be a naysayer or downplay how cool GPT4 is, but the term 'AGI' has existed for decades and is suffering from general confusion and inflation (again like 'crypto' now being read as 'coins' or even 'NFT' by the general public). See eg: https://en.wikipedia.org/wiki/Artificial_general_intelligence https://en.wikipedia.org/wiki/Artificial_general_intelligenc... or better read Bostrom's book "Super Intelligence"

wizofaus 4y ago

I've posted a number of times why I don't think current GPT tech classifies as AGI and certainly not human-like intelligence, as have others. Ask it to solve novel mathematical and logic problems, for a start (even ones you'd expect high school kids to manage), or simple word puzzles (anagrams and the like - again, avoiding "known" examples that it's likely to have come across in its training data). Further, an AGI should logically attempt to "improve" its understanding of the world during a conversation, yet chat GPT never initiates a topic or asks you anything at all (even to clarify what you mean). It also repeatedly reminds you that it is only a language model.

antibasilisk 4y ago
Machines can never obtain human like intelligence because they are foundationally not human like, that doesn't mean they lack intelligence. That's like saying dolphins lack intelligence.
- wizofaus 4y ago
  I wouldn't claim ChatGPT lacks intelligence though. Clearly in terms of language comprehension and generation it does so.
- z3c0 4y ago
  False equivalence. Dolphins have exhibited general problem-solving skills in a myriad of situations, and thus qualify as a general intelligence.
  - wizofaus 4y ago
    Indeed, I had started to write something similar. And pretty clearly animals like Dolphins or Chimps when they exercise their general intelligence don't do it by manipulation of language tokens. BTW I tried to get Chat-GPT to solve a problem chimpanzees have been observed being able to solve (using an inverted bowl to stand on in order to reach some bananas): "I'm sorry, but it is not biologically plausible for a chimpanzee to use a bowl of water to access bananas on a tree"
    - dave84 4y ago
      GPT-4 had no issues answering this.
      - wizofaus 4y ago
        Presumably because it's similar enough to descriptions of problems and how they were solved that are in its training data.
  - antibasilisk 4y ago
    As has PaLM-E
- pixl97 4y ago
  I believe you're correct, but for possibly the wrong reasons. I believe that AI will be able to simulate a lot (if not most) humanlike action in the near future. The particular problem here is they will also simulate far more non-humanlike intelligence making the alignment issue an even larger problem than it already is.
ekleraki 4y ago
> Ask it to solve novel mathematical and logic problems How exactly do you define novel tasks and problems? My brother has had it do his programming homework (he's a teenager), written in Greek mind you and therefore novel task to the first-degree, and it succeeded. Thus it solved a logic task [1]. I have had ChatGPT parse math I had written in LaTeX and reach a correct result, do modifications and so on. > Further, an AGI should logically attempt to "improve" its understanding of the world during a conversation, yet chat GPT never initiates a topic or asks you anything at all (even to clarify what you mean). It also repeatedly reminds you that it is only a language model. Anecdata: I asked Bing (GPT4) to answer a few stuff and it asked for clarification twice in a row. The fact that it repeatedly clarifies its prompt is not a limitation of the underlying model as much as it is a limitation of the interface that injects a prompt. Dan and all the other jailbreaks out there bypass that. [1] https://en.wikipedia.org/wiki/Curry%E2%80%93Howard_correspondence https://en.wikipedia.org/wiki/Curry%E2%80%93Howard_correspon...
- UncleEntity 4y ago
  > I have had ChatGPT parse math I had written in LaTeX and reach a correct result, do modifications and so on. Interesting, didn’t know it could do that. Wondering if it could also follow along with the lemmas in various papers and reach reasonable conclusions.
  - PartiallyTyped 4y ago
    It depends, I have used Copilot to modify LaTeX and write segments of my thesis, and it works surprisingly well, though its arguments are not very precise.
- wizofaus 4y ago
  https://medium.com/illumination/i-put-chatgpt-to-the-test-heres-what-it-couldn-t-handle-5897de20481d https://medium.com/illumination/i-put-chatgpt-to-the-test-he... has some examples of simple maths/logic problems it can't really solve. It'd be good to know how much better chatgpt-4 is at them.

Karunamon 4y ago

One major reason it doesn't qualify is that it is impossible for it to learn from its input. It's right there in the name: generative pre-trained transformer GPT basically has two major modes of operation right now, training, which is done by openai, and output, which is what you are doing with the API and with ChatGPT. The model does not update its "brain", its parameters and weights, based on the conversation you have with it. Even the "conversation" presented by their chat UI is a little misleading, what's happening is that the entire conversation is being replayed into the model every time you hit send. There is also a hidden input provided by openai, responsible for the "I am a large language model" stuff, anything said early in the conversation can constrain the later output, sometimes to the point of uselessness.

PartiallyTyped 4y ago
> One major reason it doesn't qualify is that it is impossible for it to learn from its input. It's right there in the name: generative pre-trained transformer Both chatGPT and GPT models exhibit in-context learning, just as you do. The difference is that during your context updates modifications to your synapses occur as a consequence of your brain's activations. Let's entertain a hypothetical scenario where I am a 5D creature, and you are (3 Space + 1 Time)D human. I could go back to any point in your time and have any conversation that I wish with you. Each time I do that, I am effectively resetting your brain's state to that particular point in time. The analog here is that for ChatGPT and the likes, time does not exist in the manner that it does for us, they are purely abstract in that they don't even need physical form to exist, it's a collection of ones and zeros that we store in such form because it is convenient. You can very well store the parameters of the GPT models in any other medium capable of holding information and actually execute the model. You can even compute gradient updates, or hook the models up to external data sources. In the same vein, you "could" do the same for any human brain no matter how computationally prohibitive it may be. You'd still classify the human brain as a general intelligence. Furthermore, we classify humans with long-term amnesia or inability to form new memories generally intelligent even though they only exhibit in-context learning.
- Karunamon 4y ago