The slide held four names in a column built for one.

I was sitting at the back of a quarterly review, a few years into my agile coaching work, watching a steering committee walk the room through the year's largest bet. The RACI chart was clean everywhere it did not matter. Responsible had names on it, so did Consulted, and Informed carried a small crowd. The Accountable column, the one that by its own definition holds a single person who answers for the result, had been filled with the name of a department. I asked who owned the outcome. Four people answered at the same time. That was the answer.

That meeting comes back to me every time I read this year's numbers on who owns artificial intelligence inside companies. LHH asked senior leaders who is accountable for AI strategy. Thirty-two percent named the C-suite as a group, twenty-two percent pointed one level down, twenty-seven percent landed on individual business leaders, and seventeen percent handed it to functional heads in HR, finance, and legal.1 Sit with that spread for a second. No answer reaches a majority. None of them clears a third by much. The largest bet most of these organizations will place this decade has been spread across the chart so evenly that it comes to rest on no one's desk.

It gets stranger one level up. When Pearl Meyer put the same kind of question to boards and executives, the two groups did not match. Boards largely assumed the C-suite owned AI, and the C-suite did not agree that it did.2 The same survey found every director believed the senior team worked as one unit, while only two-thirds of the executives on that team thought so. The people responsible for assigning ownership cannot see the gap where ownership was supposed to go.

Why the column stays empty

A blank Accountable column is easy to read as an oversight. It almost never is. Leaving that field empty is a decision, and usually a sound one for the person making it.

To own the AI mandate is to own its failure, and the odds of failure are no secret. Most of the money poured into generative AI has returned nothing measurable, with one widely cited study putting the share of organizations seeing no real return at ninety-five percent.3 Put yourself in the chair of a senior leader who has watched three change programs get rebranded and shelved. The protective move is to make sure that when the mandate stalls, the stall has no single address. A committee cannot be fired, and a working group does not get its variable comp cut. Distributed accountability feels like collaboration from the inside, and it holds up beautifully right until something actually has to happen.

What an agile coach sees

I spent years as a Navy intelligence analyst before I ever coached a team, and then more years inside groups trying to get software shipped, and the pattern holds in both places. Work moves at the speed of whoever is allowed to decide. When you cannot point to one person who owns a result and carries the authority to change course, the work falls back on consensus, and consensus is where ownership goes to die.

There is a progression I lean on, named for a colleague, that runs from compliance, through commitment, to ownership. Teams begin by doing what they are told. Given the right conditions, they start to commit to outcomes they helped shape. Eventually, when authority and accountability sit in the same set of hands, they own the thing outright. You cannot skip to the end of that progression. You certainly cannot reach ownership of a mandate that was never handed to a single owner in the first place.

The question is never whether your people are accountable. It is whether anyone was actually given something to be accountable for.

Naming the A

The fix is unglamorous, which is most of why it gets avoided. Put one name in the column. Give that person decision rights that match the responsibility, including the authority to kill the parts that are not working without convening a tribunal first. Let the rest of the chart stay a chart. What the organization needs is not another alignment meeting about AI. It needs a person who wakes up knowing the result is theirs, and a board that knows exactly whose door to knock on when the quarter turns.

I have watched the empty-column habit survive reorg after reorg, because it shelters everyone who benefits from it. The cost lands somewhere else, on the strategy that never quite executes and the bet that never quite pays.

So here is the question I would put to any leadership team reading this. If the AI initiative fails next year, do you already know the single name that failure belongs to? If you have to check a slide to answer, you have found the real problem.