Norman The AI Psychopath: They Tried To Warn UsDamien Moye
So a Facebook friend sent me a picture of McDonalds. But here, there are no clerks, just AI stands taking orders. It grieved him, and me also. However, something more disturbing is on the rise. Norman the AI Psychopath: They tried to warn us.
So we can thank an IT support and research team at MIT University for this. They literally named it Norman, after Norman Bates from Psycho. This also refer to this AI technology as ‘psychopath’. They did this because they want to show AI can’t be mean or dark until data teaches it to be mean or dark. Then they tested Norman by feeding it ‘darkest corner of Reddit’.
For example, they typed in two messages. One was, “a group of birds sitting on top of a tree branch”. But Norman saw, “a man is electrocuted and catches fire to death”. Then they typed, “a black and white photo of a baseball glove”. But Norman saw it as, “man is murdered by machine gun in broad daylight”. This was after they put the darkest corner of Reddit in Norman. So some would say, “It’s the person controlling AI. That’s the issue”. There is some truth to that. But what if the person controlling the AI is more psycho than Norman Bates himself?
The late Stephen Hawkings said AI could be the worst thing in the world. Elon Musk said AI can be more dangerous than nuclear weapons. In fact, he wants colonies on Mars and the moon so people can escape AI. Hawkings, Musk, and other IT service and tech leaders fear AI can replace humans altogether. But this test proves that humans can make AI assume and think the worst. That means people can also train it to think the worst. Then keep in mind AI can double it’s own knowledge several times over. So imagine what it can train itself to do. Norman the AI Psychopath: They tried to warn us. Are we listening?