In recent totally terrifying and not-Black-Mirror-related news, it has been found that artificial neurons, or in much more cooler terminology “superconducting computing chips” modelled after neurons have been found to be able to process information much faster and more efficiently than the human brain. So what does this mean for us and the software that we use?
Most interestingly, as these chips are able to mimic neural processing they will be able to increase the efficiency of artificial intelligence, meaning that for AI fanatics, robots just got a whole more human. Whether this information terrifies you or just merely induces you to pack your bags and move to an island untouched by robots, this news is incredibly exciting. Despite the forward-thinking nature of the sci-fi realm, artificial intelligence has often stumbled when trying to create more “life-like” intelligence.
This is most often determined by the Turing test, developed by Alan Turing in 1950, in which participants had to determine if they were talking to a machine or a human. Machines that pass this test are said to possess some degree of artificial intelligence. However, this development in AI technology marks a key moment in the development of advanced computing devices designed to mimic human behaviour.
Image from the film ‘Ex Machina’ Source: flickr
Artifical intelligence has recently received a huge amount of commercial interest, with movies like Ex Machina reflecting the publics obsession with intelligence machines. Most recently, a robot called “Sophia” developed by Hong-Kong based company Hanson Robotics found fame as the first robot to receive citizenship from any country (in this case Saudi Arabia). Sophia is able to display up to 62 different facial expressions, and her international acclaim saw her grace the covers of fashion magazines, only proving further the public interest in such technology.
The software itself is also steadily improving by getting closer to imitating the numerous functions of the brain, like Google’s automatic image-classification and language-learning programs, which use vast networks of artificial neurons to perform complex tasks such as classifying thousands of images into categories and detecting individual objects and faces.
The former has been incredibly effective commercially when used in social media platforms, and to moderate offensive content online for children. However, as conventional computer hardware is not designed to run brain-like algorithms, this complex software requires far more computer power than the average human brain does.
‘these chips are able to mimic neural processing’
Arguably this lack of efficiency has lead scientists to question if there is a better method to imitating brain processing power and has lead to the emergence of many groups invested in pursuing this research further. Many of these groups have been trying to develop ‘neuromorphic’ hardware that mimics the brain in hope that this will allow us to run artificially created neuron software better.
A key aspect of this research is the study of how the morphology of individual neurons and circuits affect how information is presented and incorporates learning and development, the latter being key in AI development. If this research is successful it could improve the quality of current robot prototypes, and potentially could lead to autonomous machines.
Sophia speaking at the AI for GOOD Global Summit, June 2017. Source:wiki commons
Despite these revelations, it is clear that the field of artificial intelligence has a long way to go before we can even hope for the possibility of a world in which AI becomes fully integrated into our daily life. The ability to replicate a natural wonder such as the human brain, which is made up of roughly 86 billion neurons, each of which transmit 1,000 nerve impulses per second, is a feat of human nature which may take generations of scientists to reproduce.
