Exploring the Feasibility of ChatGPT for Improving the Quality of Ansible Scripts in Edge-Cloud Infrastructures Through Code Recommendation

Edge-cloud system aims to reduce the processing time of Big data by bringing massive infrastructures closer to the source of data. Infrastructure as Code (IaC) supports the automatic deployment and management of these infrastructures through reusable code, and Ansible is the most popular IaC tool. As the quality of Ansible script directly influences the quality of Edge-cloud system, many researchers have studied improving the quality of Ansible scripts. However, there has yet to be an attempt to leverage the power of ChatGPT. Thus, we study to explore the feasibility of ChatGPT to improve the quality of Ansible scripts. Three raters evaluate ChatGPT’s code recommendation ability on 48 code revision cases from 25 Ansible project GitHub repositories, and we analyze the rating results. As a result, we can confirm that ChatGPT can recognize and understand Ansible script. However, its ability largely depends on how to user formulates the questions. Thus, we can confirm the need for prompt engineering for ChatGPT to acquire stable code recommendation results.