“Do you love me?” How can a computer deal with real sentences?
Blog: Decision Management Community
Prof. Roger Shank just published an article with this title. He writes: “My career was in Natural Language Processing which is only one part of AI. I spent my life trying to figure out ways that computers could understand sentences and interact with people in English. Asking people what words meant didn’t help us. People can come up with answers easily enough but they are just words.” Link
“What is love? It is a feeling. It is hard to define but people don’t need to define it. We know what we are feeling most of the time. If we don’t feel anything we will not understand the simplest of words, like hunger, anger, or ambition. Unless we are trying to get a computer to be smart we don’t have to define words. But, real AI (which these days is referred to as “AGI” so the people who do AI can pretend they are doing AI) depends on explanations of what words mean (which are based on experiences has had.)
We are very far from being able to do that. Perhaps it is time to stop making everyone afraid of AI or having endless meetings about the ethics of AI.
There is no AI and there won’t be any any time soon. Counting words or choosing between moves in GO or spitting out sentences that a person wrote, does not count.”