Today in Tech Why AI might be fueling your tech debt problem
Feb 17, 2026
David Ferrucci, former IBM Watson lead and AI/NLP expert, and Gary Hoberman, Unqork co-founder and veteran CIO, debate how generative AI and low-code can multiply hidden risk. They explore AI-generated code amplifying bad architecture, limits of no-code and vibe coding, governance and guardrails, and using component design to contain technical debt.
AI Snips
Chapters
Transcript
Episode notes
Architect For Future Change
- Design for flexibility up front to reduce future change costs.
- David Ferrucci uses the movable-wall versus foundation analogy to show some architectural choices are far more expensive to change.
Code Generator Led To $14M Bug
- Gary shared a 1994 code generator that built 700 apps but produced a $14M duplicate-wire bug.
- He learned generated code creates many unique snowflake apps that are impossible to fix en masse.
AI Amplifies Speed And Risk
- Generative AI can massively speed code production but often amplifies architectural mistakes.
- David Ferrucci warns AI accelerates experts' work but will create more bad, hard-to-maintain code when used without strong architects.