The Operations Bridge

Point thirteen — and yes, I know I said twelve closed this out. Consider this the honest correction of a leader who spoke too soon.

I said twelve was the end. Then I kept having the same conversation

In the weeks after I wrapped the series, I had a version of the same conversation more than once: someone handed up work that was wrong, and when I asked about it, the answer was some form of “that’s what the AI gave me.” Not a lie, exactly. But not an answer either. It made me realize the series was missing the point that sits underneath all the others, so I’m reopening it.

The tool produces the work. You still own it

Back at the intern metaphor, I said AI behaves like a hyper-literal, eager, inexperienced intern. Here’s the part I left implicit: when an intern hands you a draft, you don’t forward it with your name on it unread. You check it, because it’s going out under your judgment, not theirs. AI doesn’t change that arrangement. It just makes the draft arrive faster and look more finished than it has earned.

Accountability doesn’t disappear when you delegate to a machine. If anything it concentrates. The tool has no stake in the outcome, no relationship with the customer, and no consequence to absorb if it’s wrong. You have all three. That means the accountability that used to be shared across a team of people preparing the work now lands squarely on the person who chose to use the output.

“The AI said so” is not a defense

I’ve decided to treat that phrase the way I’d treat “the spreadsheet said so” or “that’s how the report came out.” It’s the beginning of a diagnosis, not the end of one. The spreadsheet had a formula error; the report pulled the wrong shift. The tool is never the reason a decision was good or bad. The person who trusted it without checking is.

A question for leaders

The next time someone on your team presents AI-assisted work, ask one gentle question: “Walk me through how you checked this.” Not to catch them out — to set the norm. The organizations that stay out of trouble are the ones where everyone already knows the answer to that question is expected before the work ever reaches your desk.