How Close Are We To The Uprising?

If the science fiction stories told through films and games were true 2018 would be far different than it actually is. The Mad Max canon would have the events of Fury Road occurring now, Blade Runners replicants would have begun to revolt, a corporate dystopia including the barbaric Running Man games would be we well established by now and of course the metal skeletons known as Terminators would be on a killing spree. This hugely popular science fiction series born in 1984 has spawned several sequels (many not so great) and yet its main storyline was one of the publics most shocking and resonant instances of robots gone mad. The idea of killer synthetic life forms and their AI technology gone rogue have been re-hashed over and over again, but despite the varying retellings of this story, something about the emotionless and unstoppable Arnold Schwarzenegger has burnt this into our minds. But is this simply another piece of fiction?

Deep Learning

In 2012 a revolution began in an area of computing that had been around since the 80’s. Deep Learning is the ability for artificial systems to continually refine their knowledge of a certain subject. Layers of programs examine data and then output their findings, where another program will begin. It’s like a learning relay. The model is based on the way neurons work inside the brain and has shown impressive results thanks to recent developments in computer hardware. Using this technology computers have been taught, or left unsupervised to perform tasks that range from recognising images of faces, to identifying the molecules of specific drugs. This has jumpstarted a great trend in AI as the deep learning systems and the neural networks they are connected to continue to grow at great haste.

Behind Closed Doors

The way it has been since computers as we know them have been around is that the programmers are really the ones who are in charge. They input the code and the parameters to the computer, and the computer executes them in whatever fashion it chooses. Now we are curious to see what happens when we do leave programs unsupervised, how do they choose to interact with the world and what do they learn? In 2017 Facebook set to orchestrate a bartering between two chat bots. The idea was for them to negotiate between themselves about the trade of some trivial items. The difference here was that the bots were allowed to communicate in whatever way they deemed necessary, which led to something unexpected – the two AI systems developed their own language. Using this new form of what appeared to be jibberish to actually end the discussions sometimes, the text-based messages were unreadable by humans. The experiment was shut down because Facebook was hoping the bots would eventually interact with other people, despite the media getting the impression that it was to avoid disaster. Still it is enough to give us pause.

The developments in AI although small today have huge potential, which may one day lead to the sentient robots we have seen in science fiction. Although programs such as Microsoft’s Zo and the chatbots that can hardly string sentences together seem light years away from human genocide, experts warn us not to be fooled. Doyoutrustthiscomputer.org shows tech pioneers who urge us to be careful with developing AI – because in the wrong hands judgment day could come sooner than we think.