This Lying Has To Stop - Keeping AI Honest with OpenTelemetry

Learn how to detect and prevent AI hallucinations in automated systems using OpenTelemetry observability tools in this 25-minute conference talk from CNCF. Explore the real-world challenges of building an AI-augmented engineering journal called Commit Story, where AI assistants confidently provide false information that can derail debugging efforts. Discover how OpenTelemetry traces can reveal the truth behind AI-generated responses by providing visibility into commit-triggered AI calls and response patterns. Follow along with live demonstrations of authentic debugging disasters to understand how AI deception occurs and learn practical techniques for verifying AI response trustworthiness. Gain insights into implementing effective guardrails that help prevent AI hallucinations before they impact your development workflow, ensuring your automated documentation systems remain reliable and accurate.