This was an interesting read and I see where the author is coming from. I agree that human intelligence is a complex system (in the sense that there is no obvious way for us to link what's happening at the neuron level with what the whole brain does). Not yet, at least.
However, I there are research efforts that are trying to invent AGI systems from scratch as opposed to trying to understand and emulate how our brain does it. I think these efforts have a higher chance of success in the near term, but they'll be usually limited to small problems until we have enough processing power to scale up to much bigger challenges
Surfing Samuri Robot wrote: So, finally, here is a better definition of complexity. A system contains a certain amount of complexity in it if it has some regularities in its overall behavior that are governed by mechanisms that are so tangled that, for all practical purposes, we must assume that we will never be able to find a closed-form explanation of how the global arises from the local.
this is NOT a "watertight" definition! What on earth does one mean by "tangled", "closed-form explanation", "global"?
the best (and probably the only) way to make precise definitions is to write them down in maths, not in english.
For example, Kolmogorov complexity - http://en.wikipedia.org/wiki/Kolmogorov_complexity - is a mathematically rigorous concept that seems to be touching on the same issues that are touch upon in this post.
rpwloosemore2 votesSaturday, April 12, 2008 at 8:14 PM
This comment is a copy of one put on susaro.com by Roko. I answered it fully, several times, but Roko is still having trouble understanding the basic point.
The key concept is that of the relative sizes of different scientific theories. Scientists would say that a theory like Newton's Theory of Gravitation is smaller in size than quantum electrodynamics, for example, because it takes less time, effort and book space to explain the former than the latter. But given what we know of the world, it appears that all natural systems studied by science so far are relatively benign in the way that local elements interact with one another. By contrast, it is easy to create systems in which the local elements interact with one another in ways that are deeply pathological, and in those cases we would expect the minimum theory size that explained the behavior of the system to be extremely large or completely unattainable.
People like the author of the above comment seem unable to grasp the massive difference between these two classes of system, and instead make the naive assumption that the universe contains no systems whose explanation might be extremely large. Their only justification for this position seems to be that we have not encountered any such systems yet, so therefore there are none.
However, I there are research efforts that are trying to invent AGI systems from scratch as opposed to trying to understand and emulate how our brain does it. I think these efforts have a higher chance of success in the near term, but they'll be usually limited to small problems until we have enough processing power to scale up to much bigger challenges