ScholarWorks@성균관대학교

초록

This study provides a comprehensive analysis of how individuals use large language models (LLMs) like ChatGPT for everyday financial management. Using survey data from 2170 Korean adults aged 25–59, this study examines the breadth and depth of LLM use across ten domains of personal finance, including budgeting, savings, investment, tax filing, debt, insurance, housing, fraud detection, financial literacy, and psychological support. Results indicate that 67.8 % of respondents have used LLMs for at least one financial task, and 58.7 % have engaged with them across two or more domains. About 15 % reported using LLMs for all ten financial tasks, while 32.2 % indicated that they have never used LLM for financial purposes. The most common applications were stock investment (50.3 %), savings planning (48.2 %), budget management (47.6 %), and tax filing and planning (46.5 %). Usage was significantly higher among men, younger adults, those with higher education, and full-time workers, whereas differences by income, wealth, and home ownership were not significant. Individuals most often used LLM as an on-demand tutor - seeking explanations of terms, concepts, and processes - or as a search engine to retrieve targeted information and compare financial products, though some utilized it for personalized advice or even emotional support. Overall, this study shows that LLMs are already widely used in personal finance, though adoption varies across financial tasks and demographic groups.

키워드