Harrison Dahme, Howl Staff
The genii at MIT have, again, brought us one step closer into the future: they have successfully built a chip which can simulate a neuron. It’s made up of about 400 transistors and is wired in such a way so that the flow of electrons through the circuitry replicates the flow of ions through a neuron’s ion channels. The uses are manifold – such a chip could ultimately enable communication between the brain and prosthetic limbs, and, closer to home, could further research into artificial intelligence.
The lead scientist, Chi-Sang Poon, stated that they “can tweak the parameters of the circuit to match specific ion channels… We now have a way to capture every ionic process inside a neuron.” The goal it is for the chip to be used in systems attempting to emulate the brain. Simulation of the brain has in fact always been one of the goals for artificial intelligence. Traditionally, these attempts focus on learning and image recognition. The methods currently in widespread use endeavor to emulate specific functions of a core group of “brain nodes.”
For quick processing, however, we are currently limited by the complexity of the system. This chip could help to fix that problem, as having a dedicated process by which to emulate the neuron pathways could greatly speed up computation and allow for higher levels of complexity with current hardware. There are, roughly 1011 neurons in the brain, each with 7000 connections. This means there is an order of 1014 pathways which must be simulated. That’s a ten with 14 zeros after it. Those are some big numbers.
That is, in fact, what some of the research happening at the U of T is focused on, which as far as I know is progressing swimmingly. Additionally, MIT released a statement saying that we may soon be within reach of brain emulation with current hardware. So what are the implications? Well – the dream of electric sheep is closer than ever. Once these chips are realized into a complete system, it will be as fast, or faster, at learning and visual processing than humans. But don’t worry there, John Connor. Sentience is still quite a ways off.