The Stack Overflow Podcast You need quality engineers to turn AI into ROI
44 snips
Jan 7, 2026 Pete Johnson, Field CTO for AI at MongoDB and tech veteran, dives into the evolving landscape of AI and its impact on jobs and productivity. He debunks the myth of AI as a job killer, emphasizing its role as a collaborator with human workers. They discuss crucial decisions for achieving ROI on AI, the importance of embeddings and vector search, and how to overcome the high failure rate in AI projects. Pete also explains how MongoDB's model aligns with modern AI workloads, advocating for curiosity and communication in the evolving tech landscape.
AI Snips
Chapters
Transcript
Episode notes
A 31‑Year Coding Journey
- Pete Johnson started coding in sixth grade and turned a childhood hobby into a 31+ year career in tech.
- He worked at HP, helped found HP Cloud Services, and joined MongoDB as Field CTO for AI.
How The OpenAI Study Measured LLM Value
- The OpenAI GDP Val study compared LLMs to experts across 44 occupations using task-level evaluations.
- OpenAI ran human evaluators to judge wins, ties, and produced a win-rate for each model.
Human+AI Outperforms AI Alone
- The white paper showed much larger gains when humans and AI collaborated versus AI alone.
- Human-plus-AI yielded notable cost and speed improvements in their joint evaluations.