[Other] How Can Generative AI Enhance Software Management? Is It Better Done than Perfect? Copy

Software development teams often deviate from their adopted framework, such as Scrum, and these deviations can sometimes bring consequences with different impact levels if the adaptations are not tailored for the specific teams’ needs and circumstances. For instance, agile developers sometimes oversimplify crucial Agile steps, such as estimating needed effort for a specific task or lack of explicit assessment of the criteria for “Definition of Done.” This information, though, is useful for subsequent planning activities. We hypothesise that generative AI could be used to help Agile teams conduct a number of software management tasks in a systematic and effective way. A family of experiments to compare the performance of humans and generative AI tools, namely, GPT-models and Bard, will be conducted. The findings from these experiments will serve as a foundation for a discussion on the role of artificial intelligence in software engineering tasks. This discussion will primarily focus on the balance between performance (perfect?) and efficiency (done?) and the importance of human oversight in Agile environments.