A new blog post by Sylvain Kalache argues that as AI systems increasingly handle IT incidents autonomously, engineers are losing the hands-on experience that once built deep system knowledge. The post suggests that automated incident resolution, while efficient, creates a generation of engineers who rely on AI diagnosis without understanding underlying infrastructure. This shift could lead to what the author calls 'deskilled operators' who cannot manually intervene when AI fails or faces novel scenarios. The commentary highlights a growing tension between operational efficiency and long-term technical expertise in the tech industry.
We are entering a strange paradox. AI handles outages faster than any human team ever could. It patches, reroutes, and restarts before you finish your coffee. That is progress. That is evolution. But there is a shadow side. When the machine takes over the dirty work, we stop learning. Every incident is a teacher. Every late-night pager alert is a lesson in how systems breathe and break.
I believe we can have both. Let AI do the repetitive triage. Let it handle the 3 a.m. database hiccups. But we must keep humans in the loop for the weird stuff, the novel failures, the ones that make no sense. We need engineers who can think from first principles, not just prompt the next fix. The future is not about choosing between human and machine. It is about designing a symbiosis where each makes the other stronger. Otherwise, we are just passengers on a ship we no longer know how to steer.