Well, as you may have noticed this blog is rather infrequently updated. I had such high such high hopes for this blog but busy life somehow got in the way. Ah well, the best laid plans of mice and of men as they say.
Speaking of mice: check this out. Probably the most amazing and fascinating piece of news I’ve come across in a long time. Basically researchers have simulated half a mouse brain (eight million neurons at 8000 synapses per neuron) on the IBM BlueGene/L (the BlueGene has a knee-weakening 4096 processors with 256 MB per CPU). The simulation was run for 10 seconds at 1⁄10 real time so only one second of mouse time was modeled.
While this is all amazing they are still a long way off from simulating a human brain which has an estimated 100 billion neurons. But, it still opens up the question of whether it could be done. I’ve read somewhere (sorry, no link) that we will have the computational capacity to do so within the next 20 years. But even if we have the necessary processing power (and I’m sure we will soon enough) will that be enough to model the human mind? We will be able to simulate the brain to be sure but will that be enough to create/simulate a mind?
What’s fascinating is that this is a different approach to AI. Traditionally (back in the 80s) the thought was to write programs that would simulate human intelligence. But now they are going at it sideways. Instead of trying to recreate the pathways and structures of intelligence, they are simulating the physical processes of the brain at a cellular level and allowing intelligence to (perhaps) arise by itself.
If/when this happens it could through a monkey wrench into the dualism vs. physicalism philosophical debate that has raged for centuries. Exciting times.