AI-assisted development: Common pitfalls and smarter practices
summary
Evidence shows AI increases 30–40% more generated code while rework/refactoring up to 20–30%, leaving net productivity under 10%. Recommendations: anchor team conventions into prompts, break work into smaller tasks, generate tests first (TDD mode), run duplication checks like SonarQube and architecture fitness functions, require human review at least as strict as for human-written code, teach both engineering fundamentals and prompting, and treat AI as a planner/reasoning partner not an autopilot.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:28AI impact on delivery
- 5:26Common AI pitfalls
- 6:08Over-trusting AI
- 11:51AI in architecture
- 14:23Poor AI usage effects
- 15:28AI integration practices
- 17:38Unrealistic velocity
- 22:37Wrong growth path
- 29:09Culture degradation
- 34:37Healthier AI culture
related talks
39mAI-assisted development: Common pitfalls and smarter practices
18mHumans + AI: The Future of Software Development Teams
34mAI and Code Quality: Building a Synergy with Human Intelligence
27mAI for software development: A reality check
33mAI assisted software delivery
36mEngineering Intent: Solving the AI Productivity Paradox
21mAI Code Quality Debt: What Strong Engineering Teams Do Differently
15mAI's Impact on Development, Post-Production & Support Journeys