Trust and Security in AI-Generated Documentation

This paper examines the potential of the Artificial Intelligence (AI) tools in the context of software development tasks and the generation of associated documentation. The first part of the paper surveys software professionals' perceptions of how AI impacts task completion times. The second part of the paper examines the utilization of the AI tool ClaudeMind to generate technical and non-technical documentation for a PDF management microservice. Two prompts are submitted to ClaudeMind, and the resulting documentations are evaluated for overall quality and suitability. This research is motivated by studies indicating that Large Language Model (LLM) performance can deteriorate with increasing complexity, highlighting the need to understand AI’s strengths and limitations in specific domains. This paper aims to provide insights into the current state of AI in software engineering and whether it can be used to automate tedious tasks such as writing and maintaining the documentation.