5 ms·

Personally I believe that a huge part of intelligence and superintelligence is about emergent complexity in networks which has increased along with the computat

Personally I believe that a huge part of intelligence and superintelligence is about emergent complexity in networks which has increased along with the computational power. Humans (and biological life in general) is as far as I can see pattern recognizing feedback loops. Humans have more connections in the brain and better/longer memory than other species which seems to be at least somehow fundamental to our difference from the rest. So while we don't know for sure IF this complexity is fundamental to intelligence we sure seem to be on the right track.

Smaug123 9y ago
If you add too much complexity of the wrong type, you get a star. Just making something more complex by no means guarantees making it cleverer. And beware of using "emergence" to hide the need for an actual explanation: http://lesswrong.com/lw/iv/the_futility_of_emergence/ http://lesswrong.com/lw/iv/the_futility_of_emergence/
- ThomPete 9y ago
  I am not talking about just complexity anymore than an evolutionist is talking about simple randomness in the evolution of species. Emerging in this context means incremental i.e. in small steps based on previous steps. Plants react to light they don't think about what light is. Not sure what that article has to do with what we talk about here.
  - Smaug123 9y ago
    It is in response to "So while we don't know for sure IF this complexity is fundamental to intelligence we sure seem to be on the right track." Which to me reads as "hey, if we make things more complex then they might become intelligent - who knows!" which seems extremely optimistic.
    - ThomPete 9y ago
      thats not what i said. I said if emerging complexity is at the core we are on the right track. I am willing to hear arguments against complexity at the center of this but then that should be what we argue. The star comparison doesent work in this vontext anymore than random does in evolution.
simonh 9y ago
Pattern recognition is peripheral I think. It's there and is a prerequisite, but not primary. The main thing we have that other advanced mammals don't to the same extent is the ability to build models, or internal simulations of situations, individuals, groups and systems in our heads, modify those simulations and model how it will behave in different situations or over time. It's this deductive reasoning that distinguishes us and I'm not sure that pattern recognition plays much of a role in it. Focusing on Complexity is putting the cart before the horse. Human brains are complex because that is a requirement to be able to do the things that they do, they are sophisticated systems that solve complex problems and therefore need to be complex. That doesn't mean that arbitrarily complex systems will somehow become intelligence because emergence. Consciousness, maybe, that might be an emergent property of our brains other functions, but intelligence and conciousness are not the same thing.
- ThomPete 9y ago
  Pattern recognition is fundmental to reasoning. Without pattern recognition no ability to establish separation between the experienced no way to build models or simulate. Without pattern recognition there would be only noise.
  - simonh 9y ago
    Sure, but I don't think that our modelling and reasoning abilities are superior to other primates primarily due to superior pattern recognition. Yes we couldn't function without it, but no I don't think it's the secret sauce we have as humans. It's what we do after we have resolved the patterns that makes the difference IMHO. Why do I think this? Pattern matching will only tell you this thing or situation is like this other thing or situation you have recognised before. That only allows you to implement a simple strategy re-use or random variation algorithm. To go beyond that you need to reason about the ways this situation is different from previous patterns, analyse it, figure out how previous successful strategies might work or might need adjusting, or conceive a new strategy. Pattern matching is just the first phase. All animals have this to some extent. But the rest is supendoudly more advanced in humans than any other animal, if they have them at all.

Smaug123 9y ago

If you add too much complexity of the wrong type, you get a star. Just making something more complex by no means guarantees making it cleverer. And beware of using "emergence" to hide the need for an actual explanation: http://lesswrong.com/lw/iv/the_futility_of_emergence/ http://lesswrong.com/lw/iv/the_futility_of_emergence/

ThomPete 9y ago
I am not talking about just complexity anymore than an evolutionist is talking about simple randomness in the evolution of species. Emerging in this context means incremental i.e. in small steps based on previous steps. Plants react to light they don't think about what light is. Not sure what that article has to do with what we talk about here.
- Smaug123 9y ago
  It is in response to "So while we don't know for sure IF this complexity is fundamental to intelligence we sure seem to be on the right track." Which to me reads as "hey, if we make things more complex then they might become intelligent - who knows!" which seems extremely optimistic.
  - ThomPete 9y ago
    thats not what i said. I said if emerging complexity is at the core we are on the right track. I am willing to hear arguments against complexity at the center of this but then that should be what we argue. The star comparison doesent work in this vontext anymore than random does in evolution.

simonh 9y ago

Pattern recognition is peripheral I think. It's there and is a prerequisite, but not primary. The main thing we have that other advanced mammals don't to the same extent is the ability to build models, or internal simulations of situations, individuals, groups and systems in our heads, modify those simulations and model how it will behave in different situations or over time. It's this deductive reasoning that distinguishes us and I'm not sure that pattern recognition plays much of a role in it. Focusing on Complexity is putting the cart before the horse. Human brains are complex because that is a requirement to be able to do the things that they do, they are sophisticated systems that solve complex problems and therefore need to be complex. That doesn't mean that arbitrarily complex systems will somehow become intelligence because emergence. Consciousness, maybe, that might be an emergent property of our brains other functions, but intelligence and conciousness are not the same thing.

ThomPete 9y ago
Pattern recognition is fundmental to reasoning. Without pattern recognition no ability to establish separation between the experienced no way to build models or simulate. Without pattern recognition there would be only noise.
- simonh 9y ago
  Sure, but I don't think that our modelling and reasoning abilities are superior to other primates primarily due to superior pattern recognition. Yes we couldn't function without it, but no I don't think it's the secret sauce we have as humans. It's what we do after we have resolved the patterns that makes the difference IMHO. Why do I think this? Pattern matching will only tell you this thing or situation is like this other thing or situation you have recognised before. That only allows you to implement a simple strategy re-use or random variation algorithm. To go beyond that you need to reason about the ways this situation is different from previous patterns, analyse it, figure out how previous successful strategies might work or might need adjusting, or conceive a new strategy. Pattern matching is just the first phase. All animals have this to some extent. But the rest is supendoudly more advanced in humans than any other animal, if they have them at all.