4 ms·

>That doesn’t require believing in sci-fi; it just requires believing in straight lines on a graph. It also requires believing the made-up labels you wrote on

by Imnimo 2y ago

>That doesn’t require believing in sci-fi; it just requires believing in straight lines on a graph. It also requires believing the made-up labels you wrote on the y-axis.

throwAGIway 2y ago
I truly don't think it does. AGI - a human level generally intelligent agent - is possible with today's technology. It'd take a lot of classical programming work - definitely man-centuries, but even if nothing new comes out and microchips stop progressing, it's possible. Look what can be done with the dumbest system imaginable - just raw input to raw output based on unfiltered conversations of internet idiots. I just can't see how it isn't possible if somebody invested the necessary man-centuries, which isn't that much - a mid level software house with few years of time.
- mdp2021 2y ago
  You have thrown in your opinion and not countered the content of the poster. That «y-axis» mentioned states that GPT-4 would be at the level of some «Smart High Schooler». I have been unable to be focused on LLMs in the past months, but unless the LLM revises its internal body of knowledge - which, if it happened, should be quite well advertised - it is not at the level of any relevant minor animal brain.
  - throwAGIway 2y ago
    I don't want to counter it. They have their opinion and that's fine, I don't agree nor disagree. I'm trying to say that we don't need any further advances in the science of ML or computers - we could already do it with a monumental lot of engineering. ChatGPT is what I meant by saying "the dumbest system - raw input to raw output based on conversations". Several engineering steps are necessary IMHO: - go through the training data and clean it up - likely what OpenAI does right now, but it's going to take a lot more than they have done so far - make it a complex system with thoughts, memories, reasoning loops, planning, goal setting, self introspection... - this is what many amateurs are trying but the professionals seem to think their time is better used elsewhere right now - careful machine code-level optimization of the software for purpose-built hardware So, no new technologies or science are strictly necessary - "just" man-centuries of engineering. Right now companies and researchers seem to think that there are more gains to be made at the fundamental scientific/technological level. But if that doesn't pan out, it doesn't mean AGI will never exist - just perhaps later, more energy consuming and less powerful than they thought. Just imagine how dumb you would look if everybody heard the first thought your brain came up with, you couldn't stop yourself from saying it, think about it to consider the implications and meaning - or at least be aware you're saying it... I'm a sleep talker and the bullshit I say without any recollection of it is just insane, and frankly very similar to a hallucinating ML model.
    - mdp2021 2y ago
      > Just imagine In fact, there is no question about revision of ideas being crucial. Some seem to ignore it. Others do not know exactly how to implement it.
  - kubanczyk 2y ago
    > it is not at the level of any relevant minor animal brain It's not at the level of any irrelevant animal brain. An LLM doesn't have to compete with mice. It doesn't even have to compete with ants. It's in symbiosis with human civilization, it's consuming our electricity and we consume it's output.
- throwaway290 2y ago
  There is no evidence "AGI" is possible at all even in theory let alone with today's technology.

by Imnimo 2y ago

>That doesn’t require believing in sci-fi; it just requires believing in straight lines on a graph. It also requires believing the made-up labels you wrote on the y-axis.

throwAGIway 2y ago

I truly don't think it does. AGI - a human level generally intelligent agent - is possible with today's technology. It'd take a lot of classical programming work - definitely man-centuries, but even if nothing new comes out and microchips stop progressing, it's possible. Look what can be done with the dumbest system imaginable - just raw input to raw output based on unfiltered conversations of internet idiots. I just can't see how it isn't possible if somebody invested the necessary man-centuries, which isn't that much - a mid level software house with few years of time.

mdp2021 2y ago
You have thrown in your opinion and not countered the content of the poster. That «y-axis» mentioned states that GPT-4 would be at the level of some «Smart High Schooler». I have been unable to be focused on LLMs in the past months, but unless the LLM revises its internal body of knowledge - which, if it happened, should be quite well advertised - it is not at the level of any relevant minor animal brain.
- throwAGIway 2y ago
  I don't want to counter it. They have their opinion and that's fine, I don't agree nor disagree. I'm trying to say that we don't need any further advances in the science of ML or computers - we could already do it with a monumental lot of engineering. ChatGPT is what I meant by saying "the dumbest system - raw input to raw output based on conversations". Several engineering steps are necessary IMHO: - go through the training data and clean it up - likely what OpenAI does right now, but it's going to take a lot more than they have done so far - make it a complex system with thoughts, memories, reasoning loops, planning, goal setting, self introspection... - this is what many amateurs are trying but the professionals seem to think their time is better used elsewhere right now - careful machine code-level optimization of the software for purpose-built hardware So, no new technologies or science are strictly necessary - "just" man-centuries of engineering. Right now companies and researchers seem to think that there are more gains to be made at the fundamental scientific/technological level. But if that doesn't pan out, it doesn't mean AGI will never exist - just perhaps later, more energy consuming and less powerful than they thought. Just imagine how dumb you would look if everybody heard the first thought your brain came up with, you couldn't stop yourself from saying it, think about it to consider the implications and meaning - or at least be aware you're saying it... I'm a sleep talker and the bullshit I say without any recollection of it is just insane, and frankly very similar to a hallucinating ML model.
  - mdp2021 2y ago
    > Just imagine In fact, there is no question about revision of ideas being crucial. Some seem to ignore it. Others do not know exactly how to implement it.
- kubanczyk 2y ago
  > it is not at the level of any relevant minor animal brain It's not at the level of any irrelevant animal brain. An LLM doesn't have to compete with mice. It doesn't even have to compete with ants. It's in symbiosis with human civilization, it's consuming our electricity and we consume it's output.
throwaway290 2y ago
There is no evidence "AGI" is possible at all even in theory let alone with today's technology.