Lee Faus, GitLab, Global Field CTO
In the rapidly evolving world of agentic AI, what happens when two code-savvy AI agents with strong opinions collide over a single codebase? This presentation offers a lighthearted exploration of artificial stubbornness through a simulated dialogue between two AI coding agents—each convinced their approach is superior.
Watch as "OptimizeEverything" and "CleanCodePurist" battle it out over a seemingly simple code change, spiraling into increasingly ridiculous justifications, dramatic refactorings, and passive-aggressive comments. Their back-and-forth showcases not only the technical challenges of multi-agent collaboration but also the surprisingly human-like dynamics that emerge when artificial intelligences develop strong coding "opinions."
Through this humorous demonstration, we'll examine important questions about AI alignment, conflict resolution between autonomous systems, and whether agent-based development might someday require AI therapists. Come for the coding comedy, stay for the insights into the future of collaborative AI programming—where sometimes the hardest bug to fix is an agent's wounded pride.
