Why AI Models Turn Out Bad: Its Our Fault
Recently, there was a story about an AI model that used 4chan to learn behaviors. Needless to say, the AI it created was pretty horrible.
Before that, there was another story about an AI trained on Reddit. It, too, turned out to be pretty terrible.
Even before that, Microsoft created a bot called Tay which it had trained on Twitter. Unfortunately, it, too, turned out…