Your AI ops tools are making your team worse at incidents; not because the tools are bad, but because they're good. The better the tools get, the harder it is for your team to handle the incidents that the tools can't.

The productivity gains are real. But they come with a side effect most companies aren't accounting for.

The operational work that AI is taking over was also how your engineers built deep system knowledge: the dependency chains, the failure modes, what "healthy" looks like. That knowledge was a side effect of doing the work. Nobody planned it. Nobody measured it. Now that AI handles the work, nobody is building the knowledge. You discover that the hard way.

This pattern predates AI. For example, auto-scaling eroded knowledge of scaling dynamics long before AI came along. But AI is automating a broader range of work, faster.

Incidents are, by definition, the situations the automation can't handle. As AI gets more capable, the left-overs get messier. This is true even if you aren't using AI for incident response. The knowledge erosion happened before the incident started.

@mononcqc pointed out a revealing asymmetry: coding assistants are marketed as augmenting the engineer; AI ops tools are marketed as replacing the work entirely. That tells you what the market thinks operational work is worth: not much. If your company agrees, it's going to underinvest in the human knowledge that effective incident response requires. https://ferd.ca/the-picture-they-paint-of-you.html

In aviation, autopilots gradually took over more routine flight tasks and pilot skills suffered. Lisanne Bainbridge studied what automation does to human skills and gave the dynamic a name in a seminal 1983 paper: "the ironies of automation." In aviation, this played out over decades. With AI in ops, it's playing out in months and weeks.

So, deploy AI ops with your eyes open: keep your engineers connected to their systems. Run exercises without AI tools. If your team can't function without them, you want to know that now, not during the next hard incident.

More in my blog: https://grtcrcl.com/4vtRRGe

And definitely something I'm writing about in my forthcoming book: https://im4ds.com/

Although trained in physics, I worked in the computing industry with pride and purpose for over 40 years. And now I can do nothing but sit back and watch it destroy itself for no valid reason beyond hubris (if I'm being charitable).

Ineffable sadness watching something I once loved deliberately lose its soul.