Is it really that unexpected that the first AI systems were kind of dumb and made easy mistakes? Shouldn't that have been, like, our starting assumption? The first cars that were invented were not very good cars! The first telephones were horrible.