User Misconceptions of LLM-Based Conversational Programming Assistants

Researchers analyzed user misconceptions about LLM-based programming assistants like ChatGPT, finding users often have misplaced expectations about web access, code execution, and debugging capabilities. The study examined Python programming conversations from WildChat dataset and identified the need for clearer communication of tool capabilities to prevent over-reliance and unproductive practices.