You’re afraid of AI? Bad news, it’s already here writes Marten Kaas. Good news, it’s not all science fiction type maniacal murdering machines. The philosopher of AI tells Motley what we can do with intelligent machines and what we can expect from our machine overlords. 

When you mention artificial intelligence, most people invariably imagine glowing red eyes peering out of a mechanical skull. But science fiction doesn’t always paint an accurate picture of the future. While it’s more exciting to imagine that a military defence system gains self-awareness and immediately begins waging war against humanity, the rise of the machines is probably better thought of in terms of a rising tide. Or, as Hans Moravec put it in 1998, a “Great Flood.”

Such a description also, admittedly, sounds horrifying, but artificially intelligent machines have as much potential to help humanity as much as they do to harm it. One prescient example was the use of artificial intelligence to predict the outbreak of COVID-19. That’s right, on December 31st 2019, alarms were raised by a company called BlueDot that a pneumonia of unknown cause was detected in Wuhan, China that had the potential for international spread via commercial air travel.

On the lighter side of things, artificially intelligent systems can be put to use to amaze and entertain people. AIVA (Artificial Intelligence Virtual Artist) for example is a music composition system that is able to generate musical compositions in a number of different styles according to different configurable parameters. Did you have difficulty coming up with a Halloween costume this year? Unsure what that perfect name for your new kitten is? Never fear, AI is here! Janelle Shane uses neural networks to do all sorts of things including generating costume ideas (why not dress up as a gothy giraffe or a sexy flying dutchman?) as well as names for cats (personally I’m a fan of Jexley Pickle).

As impressive as these feats are, however, it is important to remember that we are a long way away from true general artificial intelligence. While machine learning techniques have allowed certain systems to attain superhuman chess-playing abilities, for example, machines are still a long way from achieving human-like intelligence. Some humans, after all, can do more than just play board games. We can use language coherently, sensibly navigate our environment, invent and play sports, paint and sculpt and do appreciably more. True, machines are getting better and better at doing these sorts of things as well, but the catch is that a single machine cannot yet do all of these things. 

Photo by Piron Guillaume on Unsplash

But progress on flexible and generally intelligent machines is being made every day. That’s one major advantage that artificial entities have over biological entities: they evolve much faster. Consider that it took approximately twenty years for chess-playing machines to evolve from Deep Blue, the machine created by IBM that beat Garry Kasparov in 1997, to DeepMind’s AlphaZero. It’s not that the abilities of these systems evolved (both systems can beat a human in a game of chess), it’s how these systems acquired their chess-playing abilities that has evolved. AlphaZero, in contrast to a system like Deep Blue, is a learning machine. By playing millions of games of chess against itself, a system like AlphaZero can examine and remember far more game states than any human ever could. As Turing suggested more than half a century ago, the path towards AI is not through mere imitation but through the education of a child-like mind. This is precisely the change that has swept the field of AI over the last two decades and facilitated the rise of artificially intelligent systems. 

And make no mistake, intelligent machines are everywhere. Learning algorithms manipulate your social media feed to retain your engagement, decide if a credit card purchase is fraudulent or not, read CT scans and diagnose disease, create personalized advertisements, recognize your vocal patterns, and so much more. The Great Flood of AI is here and it is no longer a matter of whether they will displace humans but when they will displace humans, and whether we will be ready for a future shaped, for better or for worse, by intelligent machines.