The Joe Reis Show AI Bubbles & "Vibe Coding" Dangers. Freestyle Fridays w/ Matt Housley
Nov 14, 2025
In this engaging conversation, Matt Housley, a regular commentator on AI trends, dives into the shifting landscape of AI and vibe coding. He highlights the dangers of the current AI bubble and how production risks emerge from rapid prototyping. The duo discusses the importance of domain expertise in preventing errors, and the absence of solid governance frameworks for AI-generated apps. They also compare the AI sector's excitement to historical market bubbles while considering the balance of craftsmanship and mass production in software.
AI Snips
Chapters
Transcript
Episode notes
Legacy Knowledge Prevented Data Errors
- Lena Hall described a vibe-coded pipeline that missed a historic timestamp bug and produced incorrect data.
- The old platform had a manual conversion rule that the vibe-coded rewrite omitted, causing data errors.
Lock Dependencies And Use Trusted Scaffolds
- Establish an open-source scaffold with governance and vetted dependencies to limit AI pull-ins and supply-chain risks.
- Treat AI-generated dependency choices like container typosquatting risks and lock trusted packages.
Bubble Or Not, AI Progress Lasts
- Even if an AI bubble pops, the productivity gains and workflows developed now will persist for years.
- Speed increases but not always quality; proper use matters more than the tool itself.