I’ll also note that the humans aren’t reasoning in their sense. We use a combination of intuition and reasoning which feed into each other. Sometimes on the who

I’ll also note that the humans aren’t reasoning in their sense. We use a combination of intuition and reasoning which feed into each other. Sometimes on the whole problem, sometimes pieces. That has implications for what to expect out of reasoning tools trying to do what people do. A solution similar to us would probably train classical analyzers and NN’s side by side on many things. They could use the traditional tools for all the analyses computers can handle. Feed the source and analysis results to NN’s to train them. Then, use them to assist in harder, contextual analyses or probabilistic predictions. There’s a lot of unexplored territory, with potential gains, if we can mix intuition and reasoning in computers more like people do.