Large Language Models (LLMs) have been used to create artificial intelligence (AI) per- sonas to respond to surveys as humans would. However, LLMs have been criticized for their ability to mimic human behavior. The purpose of this paper is to examine the arguments that make the use of LLM as alternatives to humans less favorable. In particular, this paper examines the differences between LLM-generated and human-generated responses. The paper concludes with an overview of the current literature on LLM applications in survey research