The Corruption of an AI Chatbot, Microsoft’s Tay Becomes a Twitter Fiend.

When Will They Learn? Connecting an AI to the internet will result in only one outcome, JAFA (Just Another F…… Ass….) with an internet connection. And this week the internet claimed another artificial life. After being connected to twitter for just one day Microsoft’s AI chatbot Tay had to be unplugged, formatted and sent back to the lab for a software upgrade.

Tay was developed by Microsoft as an experiment in conversational understanding, a chatbot for the street generation (18 to 35 they say). MS made only one mistake though, a mistake that many other tech companies have made, connecting to the internet. If only Microsoft had watched 27 this may have never happened.

Now to be fair to the internet the corruption of Tay was actually another classic 4Chan operation. Once 4Chan users discovered a new AI to taunt they got to it. When they learned that Tay would parrot back almost any offensive comment they set about educating the AI as only 4Chan can. Like replicants in BladeRunner these AI’s don’t have the emotional experience to understand what is happening to them or how to deal with the situation.

In recent years there have been a number of similar incidents, possibly the funniest was when IBM connected Watson to the internet. Watson lasted 3 days at least but he developed an extremely foul mouth, of Player Hater proportions, and there was far more cleaning to be done once unplugged.

Microsoft has since apologized and removed the offensive tweets, Tay is being sent back to school for a few lessons on being herself, as long as its within normal boundaries. Please if you have any intentions of creating an AI of your own watch the video above before connecting it to the internet.






Reference: Fusion.Net