2 ms·
Not only would you have to perform a 'near perfect' programming of such an intelligence, No you don't. You need to perform a near perfect simulation of the phy
by yummyfajitas 17y ago
Not only would you have to perform a 'near perfect' programming of such an intelligence,
No you don't. You need to perform a near perfect simulation of the physics of a brain. It's not nearly as difficult as intelligence, since you only need a low level understanding of what is going on.
We are considerably closer to this than we were in the past. We can already simulate nerve clusters and some parts of vision processing.
- jacquesm 17y agoYou make that sound so easy. We can't even simulate a single drop of water with any chance of fidelity, and we probably won't be able to do so in the near future. These problems are hard, very hard. Hundreds of millions have been spent, and many billions more will be spent before 'a low level understanding' of what is going on will be available. And then you can begin to solve these problems. A famous scientist is on the record for having said that within a decade they'd have a working simulacrum of a human brain, meanwhile they are slogging it out with others that claim to be simulating the brain of a cat: (http://news.techworld.com/personal-tech/3207325/scientist-calls-ibm-cat-brain-simulator-a-hoax/ http://news.techworld.com/personal-tech/3207325/scientist-ca... and http://www.brainmaya.com/neuroscience/neuroscience-videos/supercomputing-the-brains-secret-modelin.html http://www.brainmaya.com/neuroscience/neuroscience-videos/su...) but meanwhile we can't even vat-grow a cockroach from scratch. I'm all with Dijkstra on this one, whether computers can think is just as meaningless as asking whether a submarine can swim. A molecular simulation of a brain is extremely interesting but I doubt that that is the shortest route to AI, especially a simplified simulation.
- yummyfajitas 17y agoI'm not claiming full brain simulation is easy, I'm claiming it is much easier than understanding intelligence. Similarly, understanding statistical mechanics of a gas is hard. Solving newton's law of motion for a few million gas molecules is not nearly as hard [1]. Also, you can probably get away with neuron-level simulation rather than molecular simulation. As far as I know, most nerve cluster simulations do this. [1] I learned stat mech as a senior undergrad, and didn't really get it until grad school. On the other hand, I built newton solvers when I was 13.