TY - UNPB
T1 - Addressing Agile Software Project Pain Points with Generative AI, Experimenting with ChatGPT Prompt Patterns
AU - Sainio, Kari
AU - Abrahamsson, Pekka
AU - Kettunen, Petri
PY - 2024/7/2
Y1 - 2024/7/2
N2 - Context: Agile Software Project Management (ASPM) has transformed the conventional responsibilities of the software development project management, presenting not only similar paint points found in traditional project management but also introducing new ones. Moreover, the introduction of Generative AI (GenAI), particularly through services enabled by Large Language Models (LLMs), has reached a stage where these systems can respond to user queries (prompts) with answers related to challenges in ASPM.Objective: The paper presents typical pain points found in ASPM related literature overview. Prompt engineering is used as a mechanism to utilize the knowledge base of OpenAI’s ChatGPT to address the selected paint points such as ambiguous requirements, Agile process understanding, insufficient technical knowledge and unclear roles in project team. Through empirical analysis, the suite-fullness of ChatGPT in addressing ASPM tasks with tailored prompts are assessed using selected pain point scenarios.Method: The Design Science Research (DSR) approach was employed to create and evaluate artifacts, specifically prompt patterns, across selected pain point scenarios. It then describes the development and evaluation of prompt patterns using zero-shot and second-shot prompts.Results: Ten distinct prompt patterns were designed and demonstrated using ChatGPT. In addition, ChatGPT responded to these designed prompt patterns convincingly, providing analysis and recommendations for addressing selected pain points. By employing similarity checks on the responses, potential outliers in ChatGPT’s replies when using the same prompts repeatedly were not identified.Conclusion: ChatGPT, as a representative of LLMs, demonstrates a tool for software project managers and suggestions to address pain points. However, since its responses may contain misinformation, using such a tool effectively requires a solid understanding of ASPM to ensure the correct interpretation of the responses. Using prompt patterns, ChatGPT provides valuable insights and acts as a helping tool for project management in addressing pain points.
AB - Context: Agile Software Project Management (ASPM) has transformed the conventional responsibilities of the software development project management, presenting not only similar paint points found in traditional project management but also introducing new ones. Moreover, the introduction of Generative AI (GenAI), particularly through services enabled by Large Language Models (LLMs), has reached a stage where these systems can respond to user queries (prompts) with answers related to challenges in ASPM.Objective: The paper presents typical pain points found in ASPM related literature overview. Prompt engineering is used as a mechanism to utilize the knowledge base of OpenAI’s ChatGPT to address the selected paint points such as ambiguous requirements, Agile process understanding, insufficient technical knowledge and unclear roles in project team. Through empirical analysis, the suite-fullness of ChatGPT in addressing ASPM tasks with tailored prompts are assessed using selected pain point scenarios.Method: The Design Science Research (DSR) approach was employed to create and evaluate artifacts, specifically prompt patterns, across selected pain point scenarios. It then describes the development and evaluation of prompt patterns using zero-shot and second-shot prompts.Results: Ten distinct prompt patterns were designed and demonstrated using ChatGPT. In addition, ChatGPT responded to these designed prompt patterns convincingly, providing analysis and recommendations for addressing selected pain points. By employing similarity checks on the responses, potential outliers in ChatGPT’s replies when using the same prompts repeatedly were not identified.Conclusion: ChatGPT, as a representative of LLMs, demonstrates a tool for software project managers and suggestions to address pain points. However, since its responses may contain misinformation, using such a tool effectively requires a solid understanding of ASPM to ensure the correct interpretation of the responses. Using prompt patterns, ChatGPT provides valuable insights and acts as a helping tool for project management in addressing pain points.
KW - Generative AI
KW - Agile Software Project Management
KW - Prompt Engineering
KW - Pain Points
U2 - 10.2139/ssrn.4879584
DO - 10.2139/ssrn.4879584
M3 - Preprint
T3 - SSRN Electronic Journal
BT - Addressing Agile Software Project Pain Points with Generative AI, Experimenting with ChatGPT Prompt Patterns
PB - Elsevier
ER -