Member-only story
AI Coding Is a Trust-Building Game — And the Core Reason We Talk Past Each Other
Why scaled AI practices can’t be built overnight
Simon Wang7 min read·Just now--
Over the past year, I’ve watched engineers around me pick up AI coding tools. Some got excited. Some rolled their eyes. Most just quietly tried it and formed opinions they never shared.
I’ve also gone through my own journey, one that didn’t follow the trajectory I expected.
Most conversations about AI coding focus on tools. Which model, which IDE integration, which prompt. But the actual journey isn’t about tools. It’s about trust. Learning when to trust it, when to push back, and how to build that judgment yourself.
I think naming the phases helps. It explains why conversations about AI between engineers feel like people talking past each other.
The Five Phases
Phase 1: Skepticism
“This is just autocomplete with better marketing.”
You’ve heard the hype, maybe tried it a few times, got underwhelmed. The output looks plausible but needs so much fixing that it feels faster to just write it yourself. And honestly? At this stage, with this level of interaction, it often IS faster.
