Large language models trained on questionable stuff online will produce more of the same. Retrieval augmented generation is one way to get closer to truth.
This is a companion discussion topic for the original entry at https://www.infoworld.com/article/3711100/reining-in-the-bs-in-ai.html