Understanding the behavior of LLMs and the need for explanations

Users seek explanations to determine the credibility of a chatbot's responses, while regulators require explanations to establish AI safeguards. LLMs' behavior comprehension can aid in understanding human cognition. LLMs are described as stochastic parrots, combining text patterns probabilistically without true comprehension; however, some argue they display reasoning and human-like abilities beyond mere mimicry.