The Problem is that it isn't smoke and mirrors. People see legitimately novel and exiting new things computers can do. The Problems is that those things are ex

The Problem is that it isn't smoke and mirrors. People see legitimately novel and exiting new things computers can do. The Problems is that those things are extrapolated into the future based on insanely optimistic predictions. Up until now every single innovation has had times of rapid advances and adoption followed by a slowing down and stagnation. Believing that artificial neural networks aren't subject to that trend is, again, extremely optimistic. The greatest folly, I believe, is people who look at arbitrary problems and imagine AI solving them. This will not happen. Yet I see that thinking again and again. Artificial neural networks are good at certain things and bad at others. OpenAI has succeeded massive on making people believe that it can continually improve on basically every problem at an accelerating rate. If that is true, it is obviously the most valuable company in the world. But if artificial neural networks behave like any other innovation there will be a point at which implementation is impractical and improvements will be limited.