The software includes AI opponents. In the legacy Gitlman programs, the AI was programmed to be "forgiving." If you made a standard bid, the AI would assume you knew what you were doing. If you made a wild bid, the AI would try to recover. This mimics the patience of a good teacher.