Machine learning

For 35 years, Cycorp in Austin, Texas have been trying to encode a database of basic everyday knowledge as logical data computer algorithms can process. The hope was that it would enable artificial intelligence software to leap towards human-like abilities. In the mean time, “deep learning” based on massive datasets has taken over the field, and our new AI overlords don’t think they need symbolic logic, and that the whole approach failed. Might they be wrong?

One thought on “Machine learning

  1. Eventually, a massive database of common-sense everyday knowledge will be needed. Google will make a spider that assembles it from cyberspace in a trivial amount of time, and thirty-some years of effort will be duplicated in a week.

Leave a comment.