On Sat, Oct 13, 2018 at 6:12 AM John Rose <johnr...@polyplexic.com> wrote: > > It takes kT ln 2 = 9.57 x 10^-24 joules per kelvin to retrieve (and > > copy) a bit of information. > > > > Interesting! That's an average I bet. When there are many bits intelligence > would optimize the sum?
Actually, no. That is the Landauer limit, the least energy that doesn't violate the second law of thermodynamics. Any less and you could construct Maxwell's Demon. RAM in silicon uses about a billion times more energy, which is why in a few years we will need nanotechnology to continue Moore's Law. It takes much less energy to compute by moving atoms than electrons. > You can only dance around theories of consciousness for so long before they > start coming right at ya. Here is an interesting read, KT consciousness based > on K-complexity: > > https://academic.oup.com/nc/article/2017/1/nix019/4470874 I am dubious of papers without a section titled "experimental results". I mean, it seems plausible to say that consciousness is correlated with ability to compress perceptual information, which is equivalent to saying it is correlated with intelligence. But so what? Perhaps instead of starting with the axiom "there exists consciousness", start with "human brains believe that they have subjective consciousness". The former as an axiom leads to all sorts of absurd conclusions when you start doing thought experiments about copying brains. Sort of like adding the axiom "1 = 2" and then proving that all of mathematics is inconsistent. The latter removes the mystery. It becomes simple to show that all sorts of simple programs can have the same beliefs about consciousness as humans. For example, printf("I have subjective consciousness\n"); -- -- Matt Mahoney, mattmahone...@gmail.com ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T55454c75265cabe2-M7ed374b1412fb99cd375bf90 Delivery options: https://agi.topicbox.com/groups/agi/subscription
