Does AI affect your ability to troubleshoot Kubernetes?
retained text
In my position, I often do screening interviews with software engineers for a K8s-based project. I don't really do a deep dive. Simple questions like: why do we have K8s, what is K8s, what are the essential components, etc. Among broader topics, I ask candidates to walk me through troubleshooting a failing application running in a K8s cluster. What I started noticing recently is that someone who had seemed really competent up to that point may suddenly struggle to explain even a basic troubleshooting path. One candidate was honest enough to say: *I just use an AI agent to troubleshoot this for me.* This made me wonder: do you feel you're losing some of the skills you had before the AI era? And a follow-up question: should interviews change and adapt to the AI era? I mean, these days I wouldn't expect someone to implement a controller from scratch by hand, and it's probably fine not to keep every implementation detail in your head. But troubleshooting feels different to me. Is it now also fine to just rely on AI? Or should an engineer still be able to reason through the problem without it?
document fetched 2026-09-19 00:36 · canonical https://reddit.com/r/kubernetes/comments/1wjuo8g/does_ai_affect_your_ability_to_troubleshoot