Human-AI Collaboration in Agile Environments: Survey Results

The study examined how 73 Agile practitioners collaborate with AI across different software development activities, both today and in the future. For each activity, practitioners were asked to indicate the human-AI collaboration mode they currently use and the mode they expect to use in the future. The Human-AI collaboration modes were: No AI Usage: I don't use generative AI tools AI as Assistant: I use AI for specific tasks like code suggestions or documentation AI as Collaborator: I work alongside AI, iterating and refining together AI as Lead: AI generates most of the work, I review and approve Fully autonomous: AI completes work independently without my review The software activities explored were: Requirements Gathering: Understanding what users need and what problems the software should solve Requirements Management: Organising and tracking all requirements to ensure nothing gets lost and priorities stay clear Planning & Estimation: Figuring out how much work is involved and creating a realistic timeline for delivery User Interface Design: Designing how the software looks and how users interact with it System Architecture Design: Planning the technical structure and components that make up the software Implementation: Building the actual software by writing and integrating code Quality Assurance & Testing: Testing the software to find and fix bugs before users see them This repository contains: The complete survey design, including all questions. Downloaded from Microsoft Forms as a PDF. The anonymised quantitative data collected, such as demographics and selected Human-AI collaboration modes per software activity.