AI & Systems Debt
If there's one AI article you read today, make sure it's this one: https://www.forbes.com/sites/jasonwingard/2026/04/23/vibe-coding-will-break-your-company
This isn't an anti-AI or AI-is-terrible take. This is a real & grounded take on the very real problem that AI is allowing us to move faster that the decision-time is compressed.
Over the course of my career, I learned that the shift towards "seniority" isn't just that you have more of the core skills and are faster and better. It's that you become wiser about when to use those skills. You ask fundamental questions like: What new problems do we create by solving this? If we build this, who will maintain it? If we build this, will this distract us from our priorities? If we build this will we set some new expectations that we'll have to uphold? Just because something has value doesn't mean it doesn't create a form of debt in the long run. And these were all important to consider before GenAI and coding agents came along.
AI can move so fast that you can proliferate and accumulate debt faster than you realize, and faster than you can support, and faster than you may want.
We used to talk about "failing fast" (build a small MVP, try it out, learn and refine from the failures quickly, and then iterate). With AI, it's possible to flood the system with so many possible solutions that we'll further constrain our decisioning.
It all reminds me of a joke I used to make in the earlier 2010s when RSS & News Feeds had proliferated so that actual news wasn't keep pace with our ability to consume it. I'd catch up on all my feeds and then have to wait for something to happen. If today's "news" feed is any example of how much noise we can generate, as engineering leaders we need to ensure we expand our decision-time and not compress it further.