Martin Fowler and Kent Beck discussed technical debt, cognitive debt, and intent debt at the Pragmatic Summit, comparing AI's rise to past tech shifts. They emphasized test-driven development (TDD) as a safeguard against AI's unchecked code generation. Bryan Cantrill highlighted how laziness—as a programmer virtue—drives efficient abstraction, but warned that large language models (LLMs) lack this trait. LLMs prioritize output volume over optimization, risking bloated, unmaintainable systems.

Fowler noted brogrammers' vanity metrics (e.g., 37,000 lines of code daily) exacerbate this problem, as LLMs enable prolific but poorly designed output without human accountability.