The Kind of Project To Watch
Blue Brain is an ambitious attempt to model brains. It may fail, but my guess is that within a half century some project like it will success so spectacularly as to completely remake society, and turn well-placed pioneers into multi-trillionaires. From Seed:
In the basement of a university in Lausanne, Switzerland sit four black boxes, each about the size of a refrigerator, and filled with 2,000 IBM microchips stacked in repeating rows. Together they form the processing core of a machine that can handle 22.8 trillion operations per second. … This is Blue Brain. …
When he launched the project in the summer of 2005, as a joint venture with IBM, there was still no shortage of skepticism. Scientists criticized the project as an expensive pipedream, a blatant waste of money and talent. … The first phase of the project – “the feasibility phase” – is coming to a close. The skeptics, for the most part, have been proven wrong. It took less than two years for the Blue Brain supercomputer to accurately simulate a neocortical column, which is a tiny slice of brain containing approximately 10,000 neurons, with about 30 million synaptic connections between them. …
“What is holding us back now are the computers.” … In order to accurately simulate the trillion synapses in the human brain, you’d need to be able to process about 500 petabytes of data … That’s about 200 times more information than is stored on all of Google’s servers. … But if computing speeds continue to develop at their current exponential pace, and energy efficiency improves, Markram believes that he’ll be able to model a complete human brain on a single machine in ten years or less. …
Once the team is able to model a complete rat brain – that should happen in the next two years – Markram will download the simulation into a robotic rat, so that the brain has a body. He’s already talking to a Japanese company about constructing the mechanical animal. “The only way to really know what the model is capable of is to give it legs,” he says.
Anders Sandberg tells me this is actually the most impressive project to date. Details from the whitepaper:
The so far (2006) largest simulation of a full Hodgkin-Huxley neuron network was performed on the IBM Watson Research Blue Gene supercomputer using the simulator SPLIT … It was a model of cortical minicolumns, consisting of 22 million 6-compartment neurons with 11 billion synapses, with spatial delays corresponding to a 16 cm^2 cortex surface and a simulation length of one second real-time. Most of the computational load was due to the synapses, each holding 3 state variables. The overall nominal computational capacity used was 11.5 TFLOPS, giving 0.5 MFLOPS per neuron or 1045 FLOPS per synapse. Simulating one second of neural activity took 5942 s. The simulation showed linear scaling in performance with the number of processors up to 4096 but began to show some (23%) overhead for 8192 processors.
While I don't think that throwing mips at biophysically detailed models of natural brains is a particularly elegant solution for practical AI, it could still help the understanding of how natural brains work. This understanding might help in the construction of equally powerful but less ressource-hungry models.In the end it will be the ones who simulate the relevant mechanisms at the "right" level of abstraction that will give the most bangs per buck, but projects like this might provide valuable insights what more abstract brain models could look like.
Michael, I think there is a very real danger of emerging technologies such as AI giving certain groups of people huge amounts of wealth and power. Even if that power is granted through voluntary exchange, it can still be abused. People in power (i.e., in control of vast AIs) might not have many incentives to engage to trade with normal, less-powerful humans when other options (conquest of a sort?) are available to them. Trading with humans might become pointless for anything but manual labor if AIs became too intelligent.
I'm mostly afraid of what would happen if AIs of human intelligence or greater were created their distribution significantly restricted through regulation, patents, or outright nationalization. Instead of becoming a technology for the masses, AI could be a wielded by those already in power, who have few incentives to share the technology with others.
Can you imagine the anti-AI rhetoric politicians might spout in the future? I wonder if they'd think it unsafe to use for their own (Skynet?) purposes.