The Stack Overflow Podcast Professional skepticism is a dev’s best skill
12 snips
Sep 25, 2026 
David Burns, BrowserStack’s Head of Developer Advocacy and Open Source, explores why professional skepticism matters more in an AI-driven world. He discusses test-driven development for AI agents, guardrails for vibe coding, security risks, flaky tests and dirty application state, plus risk-based testing and the growing value of human QA judgment.
AI Snips
Chapters
Transcript
Episode notes
Why QA Skepticism Matters For Every Engineer
- QA thinking combines empathy with systematic skepticism, exposing problems happy-path development misses.
- Testers notice slow loading, nonsensical architecture, and workflows that fail users rather than merely checking whether buttons work.
Verify AI Output Beyond The Happy Path
- Treat every AI-generated result as untrusted until you verify it from a different perspective.
- David Burns caught a Claude Code login failure that redirected users to localhost, illustrating why someone must inspect deployed behavior.
Use Two Commits For Agentic TDD
- Use a two-commit workflow: have the agent write tests first, then implement code that makes them pass.
- Require human approval before changing tests so implementation assumptions cannot silently redefine requirements.