Press enter or click to view image in full size DALL-E generated AI is improving but still prone to hallucination, especially in complex tasks. According to the Vectara Hallucination Leaderboard, the best leading models (OpenAI GPT-4.5, Claude 4.5) have a hallucination rate of 7–10%. Although hallucination seems something new to study , as LLMs are new as a concept, it has already been studied for years. Let me explain. You have probably met this person. They speak with total authority on a subject they barely understand. No hesitation, no caveats, no “I might be wrong.” Just pure, effortless confidence — delivered with the kind of conviction that makes you doubt your own knowledge even when you are right. That is the Dunning-Kruger effect: the cognitive bias where limited knowledge produces maximum confidence. The less you know, the less you are aware of what you do not know. The danger is not ignorance — it is ignorance that does not know it is ignorant.…