InnovationRay Kurzweil

Scientists create artificial brain with 2.3m simulated neurons

By December 11, 2012 2 Comments

One of Ray Kurzweil’s central predictions is that we will create software that emulates the human mind by reverse engineering the human brain. I think that is a sensible prediction – at some point in the not too distant future we will have the hardware, and after that it is just a question of developing the software, which will naturally follow from our continually improving understanding of how the mind works.

I’m writing this today because scientists have taken a good first step by creating an artificial brain with 2.3m simulated neurons configured in networks that resemble some of the brains own networks. Its cognitive network simulates the prefrontal cortex to handle working memory and the basal ganglia and thalmus to control movements. This brain takes input via a camera like sensor that can view an image and outputs by writing characters with a robotic arm. In the video below you can see it perform a series of tests with varying levels of success – recognising digits, recalling from memory, adding numbers and completing patterns. Like many human brains it is good at pattern recognition (in this case recognising characters) and struggles a bit with short term memory (in this case drawing a series of random numbers in the order they were shown).

This is just a small start. Everything the brain does could be performed by a computer and it is much simpler than a normal brain which has an estimated 86bn neurons. But it is a promising start. There is a long way to go from 2.3m neurons to 86bn, but I imagine getting the first 2.3m working is more challenging than the next 85.9977bn, at least in the requirement for deep insight and fundamental breakthroughs.

All of this is important because once a computer can emulate a human mind it will be able to perform many, if not all, of the functions a human can perform which will have a transformative effect on society, I think for the good.