부산대학교 도서관

The aim of this study is to compare the responses provided by career counselors to online career counseling questions with those provided by ChatGPT 3.5. We asked 1,000 identical questions to ChatGPT, and conducted text mining, keyword analysis, and sentiment analysis to determine whether the responses were similar. The keyword analysis showed that career counselors used key words such as ‘job’, ‘information’, ‘counseling’, ‘career’, and ‘self’, while ChatGPT used words such as ‘certification’, ‘management’, ‘self’, and ‘ability’, which were significantly different from each other. In the cross analysis and emotional words analysis, the career counselor used more emotional words based on human understanding and empathy than ChatGPT, and in the network analysis, which looks at the connections between words, the career counselor’s connections between words such as hope, interest, new, ability, and help clearly revealed the strategic context of the job hunting process, while the connections of the ChatGPT’s answers were less coherent. These results suggest that the quality of career counseling can be improved by understanding the differences in counseling between humans and A.I., and by combining strengths and dividing roles. The analysis of which approach offers qualitatively superior counseling is left for further research.