Since GPT-3.5 became available to the public, researchers have started to ask if the LLMS “knows” or “believes” [83, 168, 194] like humans. One of the reasons to compare it with humans [189] is that a model itself is not just the knowledge it learned, but also knows/understands the relations. In [80] the models are tested with human experts. One of such comparisons is based on the Turing test and the Helpfulness test. Passing the Turing test could already prove that a model is at least on a human expert level. Together with knowledge, the way of providing it in an empathic way with a positive sentiment is the other level of the current research [116]. This is why more work is dedicated to having not just a knowledge LLM, but also one with a specific “personality,” a model that behaves appropriately to current situation and application.