Scholarlink

This study uniquely investigates the influence of ChatGPT and persona patterns on the problem-solving process in the context of complex problems. The focus is on how these technologies are evaluated for problem analysis and how they influence problem understanding, the quantity and quality of solution approaches, and the problem solvers’ sense of uncertainty, depending on persona patterns. Using a mixed-methods approach, the study combines quantitative and qualitative research methods. In an experimental design, 99 participants used ChatGPT to create issue trees to analyze a complex home office problem (loss of productivity) - once with and once without persona patterns. Using qualitative methods, the quality of the issue trees was compared with a reference. The results show that ChatGPT can be an effective tool for supporting the problem-solving process. Although the structure of the issue trees did not differ between the groups, the use of ChatGPT significantly improved the perceived understanding of the problem, the quantity and quality of perceived possible solutions, and the subjective sense of certainty. Using persona patterns had no significant additional effect on the psychological variables. The study indicates that ChatGPT is suitable to support thought processes in the image of the “second brain” and should, therefore, be augmented as often as possible. The use of personas should be directly related to the problem context. Exploratory analyses suggest that the influence of ChatGPT could be more pronounced in participants with less prior knowledge. These findings have far-reaching implications for integrating AI into problem-solving processes in organizational contexts and open new research perspectives on the nature of AI-generated problem analyses. A small sample size and psychometric limitations restrict the study’s findings. For further research, it is recommended that the study be repeated with higher resources, recommended methodological improvements be integrated, and overall problem-solving processes with ChatGPT be evaluated in longitudinal studies.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

LLMs as Your Second Brain - How AI Will Affect the Way We Solve Problems

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Phineas Speicher,
  • Lutz Becker

摘要

This study uniquely investigates the influence of ChatGPT and persona patterns on the problem-solving process in the context of complex problems. The focus is on how these technologies are evaluated for problem analysis and how they influence problem understanding, the quantity and quality of solution approaches, and the problem solvers’ sense of uncertainty, depending on persona patterns. Using a mixed-methods approach, the study combines quantitative and qualitative research methods. In an experimental design, 99 participants used ChatGPT to create issue trees to analyze a complex home office problem (loss of productivity) - once with and once without persona patterns. Using qualitative methods, the quality of the issue trees was compared with a reference. The results show that ChatGPT can be an effective tool for supporting the problem-solving process. Although the structure of the issue trees did not differ between the groups, the use of ChatGPT significantly improved the perceived understanding of the problem, the quantity and quality of perceived possible solutions, and the subjective sense of certainty. Using persona patterns had no significant additional effect on the psychological variables. The study indicates that ChatGPT is suitable to support thought processes in the image of the “second brain” and should, therefore, be augmented as often as possible. The use of personas should be directly related to the problem context. Exploratory analyses suggest that the influence of ChatGPT could be more pronounced in participants with less prior knowledge. These findings have far-reaching implications for integrating AI into problem-solving processes in organizational contexts and open new research perspectives on the nature of AI-generated problem analyses. A small sample size and psychometric limitations restrict the study’s findings. For further research, it is recommended that the study be repeated with higher resources, recommended methodological improvements be integrated, and overall problem-solving processes with ChatGPT be evaluated in longitudinal studies.