Sunday, July 12, 2009

Memristors complete the electrical spectrum

At least until memcapacitors and meminductors come, we now have all of the basic building blocks of electronic devices. New scientist has a GREAT article about memristors. Seems that one guy (Leon Chua) figured out how this missing component a.k.a. the memristor should have been back in 1971! Turns out another guy (Stan Williams) had acceidentally stumbled on an arrangement that behaves just as Chua expected around 2000. Why is this exciting? I'll tell you in a minute. Suffice to say the effect is based on charged oxygen bubbles migrating into or out of RESISTORS! It also appears it was not within our technology realm until we got much close to today in terms of fabrication techniques (i.e. photolithography etc.)

Then we get on to Biology. It appears that a gentleman by the name of Max Di Ventra (UCSD woo hoo!) connected the behavior of a large single celled slime with the memristor. Of course it turns out that SYNAPSES are just memristors (Based on their description they operate in the frequency domain) with the "ebb and flow of potassium and sodium ions" substituting for the charged oxygen bubbles. Nature had them all along!

I did a brief stint working at the neuroscience institute working on robotic autotmata for the first robocup with segways. The institute was founded and run by Gerald Edleman who did amazing work on the structure of antibodies which he recieved a nobel prize for. Basically one of the feelings I got while there was that he valued just as much trying things out as opposed to trying to unravel them. We basically worked on building models of the brain to do limited things based on our understanding as it stood at the time. Of course we were just SIMULATING neurons in a coarse, and really digital fashion and they were able to make great strides in understanding. Now all of a sudden there's a circuit element which behaves like a neuron.

So this begs the question.....How long til AI? I certainly think this is a big step toward starting to have intelligent systems because we can just copy our brains to figure things out instead of trying to simulate what ends up just being a memristor in the first place. This certainly opens new doors, and the rapid development on this project track though I still wonder at seeing a sentient machine in my lifetime. It was only a few years ago it took a hundred million dollar machine (deep blue) to beat the best chess player at CHESS! Imagine the difficulties in programming, or really as we get closer TEACHING a computer. I think it will be a while before the technology, let alone philosophy exists for us to raise a sentient computer.

No comments: