You can make the exact same question to the same LLM and the "artificial entropy" they inject into the inference process will be enough to make up a completely

You can make the exact same question to the same LLM and the "artificial entropy" they inject into the inference process will be enough to make up a completely different response.