AI can be trained to do relatively easier tasks like content moderation for explicit material or terrorist content, but it struggles with nuanced understanding of languages and contexts that can lead to false positives.