« Penguicon & Blook | Main | Qualitatively Confused »

March 14, 2008


I wonder if they could do it faster with specialist hardware? (Not that the IBM super computer isn't special, but I mean specially designed for them),

I am fairly skeptical of attempts to build brains by simply putting lots of neurons together. The brain will not simply become alive and start acting smart. "neuron-iness" is a superficial property of brains and isn't the only thing that makes them smart. At some point the information for the goals/instincts/drives has to enter the system at witch point the builders are right back at square one. Encoding goals into a neural network might end up being harder than making an AI from scratch in C++ encoded with goals.

In conclusion, despite having "neuron-iness" in it, I don't think this project is any closer to solving the AI problem.

Intelligence is the Emergent Property of computers which have computing power at least an order of magnitude bigger than the ones we have access to currently, especially if we use them to exactly simulate the Mysterious and Hidden Secrets of the Brain. Everybody knows that.

This isn't modeling some random system of neurons, it's modeling structures specifically based on mammalian brains. There may be some emergence mysticism, but there's substance too.

If they can simulate 10,000 neurons, why not work on simulating the complete nervous system of an organism with fewer neurons than that? Over-ambitious goals like rats seem designed to keep tests out in the future and to optimize the amount of funding they ask for.

I wonder how realistic the goal is to model a complete rat brain in two years. Not only would they need a computer big enough and fast enough to hold a rat brain, they would also need a fully detailed neural map of a rat brain to put into it. Are we that close to having such a map?

If and when we are ready to do such tests, one has to think about the ethical implications. Scientists seem willing to inflict hardship on rats for the greater good, so perhaps simulating rat brains doesn't raise ethical concerns. But we should consider the possibility that the simulated rat is conscious, and that it could be in a state of pain and suffering during the simulation.

"some project like it will success so spectacularly as to completely remake society, and turn well-placed pioneers into multi-trillionaires."

Pardon me if I don't break out my pom poms. Who will control the technology, and what will they use it for? Fewer, richer people are not exactly exciting anymore.

"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."

I wonder if it would be easier and just as useful to use a virtual environment first. Robotics just adds a layer of problem that isn't necessarily useful to learn about brain.

"I wonder if they could do it faster with specialist hardware? (Not that the IBM super computer isn't special, but I mean specially designed for them)"

Faster, and much more expensive.

" Fewer, richer people are not exactly exciting anymore. "

If someone makes a lot of money without stealing it (via the threat of violence or fraud, or via legal taxation and political connections), that means that the person has created value that others were willing to pay for. I want to live in a world with as many of these people as possible, because the only thing they're taking away from me is what I'm willing to give them (unlike governments), and and these technologies have a chance of greatly improving my life.

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.

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.

The comments to this entry are closed.

Less Wrong (sister site)

May 2009

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30