The interesting thing is that I don't think I have yet heard anyone mention the term overfitting (a potential design flaw). Depending on how an "AI" model is designed, it can regurgitate it's training material.