The Stack Overflow Podcast AI can 10x developers...in creating tech debt
25 snips
Jan 23, 2026 Michael Parker, VP of Engineering at TurinTech and former Docker engineer, explores the new frontier of AI-generated tech debt. He discusses the uneven productivity of AI tools, especially in legacy systems versus new projects. The conversation highlights emerging roles like developer coaches and the importance of planning and collaboration. Parker also touches on the emotional impact of AI on developers, emphasizing the need for continuous learning and adapting to new workflows. Tune in for insights on how AI can reshape development teams and processes.
AI Snips
Chapters
Transcript
Episode notes
AI Productivity Is Not Uniform
- AI productivity gains are highly uneven across dev teams and codebases.
- Averaging metrics (like 19% slower) hides extremes where AI is either useless or transformative.
Tune The AI Factory, Not Just The Code
- Invest in roles that tune AI systems instead of fixing every generated line of code.
- Build prompts, rules files, and subagents so AI quality improves without manual rework.
Fix The Four Stages Of AI Development
- Improve four stages: planning, coding (babysitting), reviewing, and maintenance.
- Build distinct tools for each stage to reduce AI-generated tech debt over time.