Early ChatGPT User Portrait

Early research on ChatGPT user behavior reveals diverse applications and evolving interaction patterns. Studies analyze user demographics, topics of interest, and sentiment expressed during multi-turn conversations, often employing techniques like Latent Dirichlet Allocation to identify thematic trends. This work also explores ChatGPT's capabilities in specialized domains, such as assisting medical diagnoses and automating scientific text mining, highlighting both its potential and limitations in these contexts. The implications extend to education, with investigations into university policies regarding ChatGPT use and its application in automated student assessment.