Artificial intelligence mirrors the many faces of being human. Our creations can be vicious and cruel as in The Terminator or curious and compassionate as in Wall-E. They can destroy our word as in The Matrix, betray us as in Ex Machina or give us a glimpse of miraculous possibilities as in Transcendence. Before we venture too far into this new era of invention, it would be better if we knew more about ourselves and the unintended consequences of designing robots to be more like us.
Recently, Microsoft got a lesson in unexpected consequences when it released “a machine-learning chatbot designed to mimic the verbal tick of a 19-year-old American girl.” (Click) They named their machine, “Tay,” and released her upon various social media where she was to interact with others to learn “new phrases along the way.” (“Artificial Intelligence: Corrupting Microsoft’s chatbot,” The Week, April 8, 2016, pg. 18.) Tay’s creators weren’t fools. They’d developed algorithms meant to screen out mischief makers who might teach the innocent Tay words a 19 –year-old girl shouldn’t use. But the mischief makers were quick to spot a vulnerability. They gave her the command “Repeat after me.” By this device, Tay learned new words and incorporated them into her speech. Soon she was talking gutter speak with the worst of them.
The outcome of Microsoft’s chatbot experiment strikes me as funny. When computer geeks get out-geeked by internet rabble, a know-nothing like me takes a little pleasure in it. Seeing our technological overlords being taught a lesson in humility is worthy of a belly laugh. Nonetheless, the experiment shows us the dark side of artificial intelligence. Presuming to create projections of ourselves opens a Pandora’s box. We humans are complex creatures. How do we presume to create a perfect robot when we have yet to master the flaws in ourselves? The only real talent we’ve shown thus far is the ability to ignore our shortcomings. Case in point. Recently Netflix, champion of net neutrality, was exposed for reducing video speed to some customers in the interest of increasing profits. (“Byte: What’s new in tech,” The Week April 8, 2016, pg. 18.)
Our desire to use robots to aid humanity will no doubt result in extending our prejudices and serving our darker purposes. So it has been since the invention of the wheel, gun powder and nuclear power. As the French writer, François Rochefoucauld once observed, “There are but very few men clever enough to know all the mischief they do.”
