Are bad incentives to blame for AI hallucinations?

How can a chatbot be so wrong — and sound so confident in its wrongness?

Read more at TechCrunch