Neural Simulations and AI

• Neural simulations are really quite far away from good old fashioned AI.
• Instead of symbols and serial processing, it all relvolves around lots (thousands, millions, or in the full-fledged case 100 billion) neurons.
• Moreover, unlike serial GOFAI, all of those neurons are behaving at the same time.
• Of course this is a modelling problem, and thus a reasonable task for AI.
• But it's more.
• It's pretty clear that thinking is done by neurons (and synapses) in humans and any type of complex life.
• So, since humans use symbols, somehow those symbols must emerge from the neurons.
• Since thinking is done by neurons, thinking could be done by simulated neurons.
• So, if we could develop a reasonably accurate neural model of a person, we'd have a full-fledged, Turing test passing, AI.
• So how might we do that?