4 replies to this topic

[abolitionist]

Electromagnetic activity in the brain can be likened to that produced by a vacuum tube;

http://en.wikipedia....tube_amplifiers

it travels in waves due to spindle neurons, supportive glial cells, and complex firing patterns associated with learning (among many other factors);

http://technology.ne...ve-neurons.html

http://en.wikipedia....spindle_neurons

Currently, you can buy vacuum tube modeling amplifiers which mimic the tone of vacuum tube amps using digital processors. Musicians seem to prefer the real deal - the warm, fluid, dynamics of tube amplifiers though they are much more expensive and difficult to keep running (vacuum tubes are not as reliable as solid state circuitry.)

Neural signals are infinately more complex than the signal generated by a single vacuum tube within an amplifier - maybe those interested in AI should first work on making a true sounding rock guitar amplifier.

http://en.wikipedia....f_consciousness

We'll need these bodies and minds for a long time before we can even make a copy of a brain much less transfer consciousness.

Edited by abolitionist, 09 December 2007 - 08:21 PM.

[Athanasios]

Electromagnetic activity in the brain can be likened to that produced by a vacuum tube
it travels in waves due to spindle neurons, supportive glial cells, and complex firing patterns associated with learning (among many other factors);

To make an AGI you may not have to, and IMO better to not, mimic the physical structure of the brain. Those working on AGI exclusively are not recreating the brain physically but developing an intelligence based all in code. The intelligence does not have to think like us and in all probability it will not. I do not see how your analogy holds at this point.

Edited by cnorwood, 09 December 2007 - 09:05 PM.

[basho]

Those working on AGI exclusively are not recreating the brain physically but developing an intelligence based all in code. The intelligence does not have to think like us and in all probability it will not. I do not see how your analogy holds at this point.

You have to be a bit careful with terminology. The code may provide the specification, but ultimately you will still end up with a physical device with a specific configuration as a result of compiling and loading the resultant machine code instructions.

There are a couple of levels here. If you consider intelligence (and by "intelligence" I mean the combined properties of consciousness/self-awareness and general-intelligence) as an emergent process arising from the neurological substrate of a biological brain, one path to AGI is to reproduce the essential dynamic or causal properties of the brain in a non-biological computational substrate *configured* via software (source-code is just a description of a particular machine configuration). The idea is somewhat analogous to a virtual-machine and is important when you start thinking about substrate-independence. With a correctly specified and implemented VM, conscious intelligence would arise in just the same way as it would in a human-brain. An important point that takes some time to appreciate is that it would not be artificial intelligence/consciousness. The engineered substrate configuration may be considered artificial (although computationally equivalent) when compared to a human brain, but the effect is just as real as that produced by its biological counterpart.

The hard part is, we still don't have a definitive theoretical understanding of the effect we are trying to produce or at what level we need to model the brain.

[caston]

If you have to plug something into a wall socket it's not really self aware. The self awareness is patterns formed by the systems own energy producers.

Edited by caston, 14 January 2008 - 02:18 PM.

[basho]

If you have to plug something into a wall socket it's not really self aware. The self awareness is patterns formed by the systems own energy producers.

Caston, you're either a genius or else there is a little too much Lysergic acid diethylamide in your supplement mix