@DrewKadel@social.coop Nice. In my own view, it’s basically a database. The main difference from a traditional database is that it’s got the data smeared out, impossible to trace back to the origin. About things where it’s got a lot of input (training), answers have good precision. In parts where it got little or no input, it doesn’t know that it doesn’t know, but produces answers that are extrapolated or dreamed up. But this view doesn't explain the part about appearing to do interactive introspection.