When AI is a teammate: A case study of integrating AI tools into development teams

AI tools are everywhere. Should they be included in the development process? What are they good for? What are their advantages and shortcomings? What impact can they make on the final product? Which ones to use?

These are the questions we cover in this case study of using AI for the development of real-life DNA design software.

For almost a year the Institute for Systems Genetics has been using generative AI tools in the development of our web application and the first-year job review of the AI ‘teammate’ is akin to an eager,  know-it-all recent college graduate, the one that requires patience and supervision, but has the potential of becoming a productive partner.

This session provides reflections on and an evaluation of the practical uses of the AI tools for the development of DNA design software, with examples of real-life successes and disappointments of the new AI ‘teammate’ that could shorten the acclimatisation phase if a team decides to partner with this new technology.

Participant takeaways:

• The motivation behind using Generative AI as an enhancer for our team’s performance. 
• Comparing and contrasting the tools tried out: ChatGPT, Gemini, and Copilot.
• Specific tasks AI was applied for algorithm design, documentation, code generation, and error message interpretation.
• Thumbs up AI performances: all of the small, well-defined tasks.
• Thumbs-down AI implementations: jobs that require creativity, testing, diagram drawing, humility, and word puzzles.

Themes:

AI, Design Process, Development, Productivity.