The Chief AI Officer is a new role, and I do not know anyone who can describe it with confidence. I believe I am doing a version of it, so I am going to explore it in public, over a series of posts, and test what I think against what you think. I think the role is still being worked out, and better worked out with other people than announced from one chair.
I will start with the one public, specific definition I know of. The US federal government has one. A 2025 memorandum from OMB lists ten duties for an agency's Chief AI Officer: promoting adoption through governance, coordinating compliance, advising the agency head, representing the agency externally, keeping the use case inventory, running the process for high-impact AI, advising on workforce, inventorying and releasing AI code and data, guiding investment, and supporting the tracking of spend. It also lets an agency head designate an existing official for the role, and NASA's first Chief AI Officer was exactly that, an expansion of its chief data officer's role. That is one government's answer, and I do not know that a company would read it the same way. It also shows why one definition is not enough. If the title can be a second hat on an existing seat, as it was at NASA, it does not pin down a single job, and the way to find out what the role is for is to compare notes.
I cannot tell you what people generally think the role is. I have impressions, not a measure, and I would rather ask than guess. What do you picture when you hear the title?
Here is where I stand. My organisation does not have a named Chief AI Officer yet, and I do not hold the title. A good part of what I do day to day lines up with those duties anyway: helping the organisation adopt AI, shaping the architecture it runs on, and representing that work outside. I am working on the work the role describes, and my working hypothesis is that the title can cover two quite different jobs, putting AI into what an organisation builds and changing how it operates.
The next posts will follow that experience, and these are rough directions, not a plan. The two jobs under one title. What measuring adoption showed, starting with who had no access at all. Who our AI sessions reach. The platform that no named duty covers. Why a title is not a seat. What approval means when an agent writes the code. There are other threads I am not sure belong yet, and I expect some of what I say to turn out wrong.
It is one organisation, so none of it is evidence about what the role usually is. The order will change, and so will the content, depending on what you tell me. I welcome the discourse wherever it goes.
What would you want a Chief AI Officer to be responsible for, and what would you worry they would get wrong?