10 Things I Hate About AI
summary
Lists ten concrete failures: top‑down mandates that create fake buy‑in, useless vanity metrics like “percent AI code,” LLM hallucinations that are confidently wrong, and frantic tool churn (Copilot, Claude, Ko and many plugins). Cites industry signals — code churn doubled (3%→6% in 2024) and code duplication up eightfold — and measured wins like Adidas’s 20–30% team boost and 3.6 hours saved per developer per week. Recommends CI/CD guardrails, TDD, outcome metrics (DORA), and preserving core dev skills.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:12Hyped AI mandates
- 2:39Terrible AI metrics
- 3:42Confidently wrong AI
- 5:10Tool overload chaos
- 6:18Weaker coding skills
- 7:19AI slop quality crisis
- 8:46Bad code locking in
- 9:51Prompting skill pain
- 11:13Large PR problems
- 12:18AI haters frustration
- 14:06Reframing the hate
- 15:11Better AI adoption
- 17:18Measuring real outcomes
- 22:59Fighting AI slop
- 26:29Quality gates matter
- 28:55Guardrails for AI
- 30:40AI literacy skills
- 35:59What it all means
related talks
34mThriving in an AI era 🔮
39mAI-assisted development: Common pitfalls and smarter practices
18mHumans + AI: The Future of Software Development Teams
23mHuman-Driven Intelligence: Where AI & Automation Fall Short
37mAI-assisted development: Common pitfalls and smarter practices
37mThe AI Factor: The Blueprint for Intelligent Deployment
49mClean Code in the AI era
36mHow to Get AI to Stop Being Weird and Actually Be Useful