The definition of AI is very vague these days. But Bayes filters are trained on good and bad data, and then learn over time automatically by what people mark a

The definition of AI is very vague these days. But Bayes filters are trained on good and bad data, and then learn over time automatically by what people mark and unmark as spam. This is exactly what things you would call AI do. A person or team manually creating filters to block spam would have a hard time keeping up with the effectiveness of a learning Bayes model.