Artificial intelligence (AI) is becoming more and more a part of our lives. Siri, in many of our cellphones, is a form of AI. More recently, AI has been able to grab other headlines recently, with Google’s Alphago being the first ever computer to defeat a human at the complex game of Go.
While this feat is impressive, not all headlines involving AI are nearly as positive. Microsoft recently released a “chat bot,” named Tay, which would interact with humans, learn from its interactions and then further develop its conversational skills. However, in an alarming turn of events, Tay quickly turned into a genocide-advocating racist.
Microsoft has since apologized and has said they are dedicated to learning from Tay moving forward. They claimed that Tay was exploited in “a coordinated attack by a subset of people” who exploited a vulnerability in the AI.
Whether you pass off this Hitler-loving AI as the result of an attack or not, Microsoft got this one horribly, terribly wrong. This has brought many people to speculate on the safety of AI and has started debates on how wary humans should be of AI moving forward.
Computers could easily dominate humans, claim these AI naysayers. And there certainly is some truth to those concerns. If Tay was able to go from innocent teen to all out racist in less than 24 hours, what would be stopping her from deciding to kill off humans given the power to?
While any AI entrusted with human life would likely be slightly more refined than Tay, there still is the possibility of vulnerabilities resulting in similar anti-social or anti-human actions from an artificial being.
This possibility in mind, developers and AI specialists should be very careful in what powers that they give to artificial minds.
The capabilities of AI are increasingly moving into a position where they have power over human life. IBM’s Watson is moving into the healthcare industry. While Watson does not have any power to decide who lives and who dies, if a computer like Tay ever got loose in the healthcare world, I would shudder to think what would happen.
The same can be said of many other industries. Therefore the recent headlines of AI should serve as both an encouragement and a caution. Alphago’s successes prove the power and potential of Artificial intelligence while Microsoft’s Tay should warn of the drawbacks of unchecked, inhuman Artificial Intelligence and the damage that it can do.
