Leading an AI project is not something you delegate to a likeable person and then admire from a distance. The most common failure I see is exactly that: a proud leader hands the work to someone who presents well, announces it as a win, and steps back into the past while the company’s whole future rides on a project they have stopped touching. I was once invited to headline a firm’s AI kickoff, and what I held back from the leader that day is what I will say plainly here.
Why the choice itself revealed the problem
The leader had picked the person in the organisation who spoke most fluently about AI, the most eloquent, the most liked. Not the person with the most demonstrated depth. And the moment you choose for fluency over substance, you have told me something uncomfortable: that you cannot tell the difference, which usually means you have not done the work yourself. If you were impressed that someone could talk well about ChatGPT, you have probably done very little, and the project is likely going nowhere. An AI project is not a communications exercise. It is an innovation effort that has to change how the company works, its culture, and people’s habits, which is some of the hardest work there is. The person who can actually drive that is often not the most comfortable one to be around, because they tell the truth people do not want.
The KPI question that exposes everything
So I would ask the leader directly: what are your KPIs for this? What are the project’s? What standard are you holding yourself to while this runs? Because what you have actually described to me is that you assigned the entire thing to someone else and felt proud of it. But the measure that matters is your customers: are they more profitable, more efficient, given genuinely new models, shown what competitors are doing. If your real measure is whether you avoided embarrassing yourself, the project has already failed. This is the same accountability gap that turns AI output into workslop nobody owns.
Your future is not something you hand off
This has to involve you, the leader, every single day. Not running it, but in it, side by side, because it is your future on the line, not the project owner’s. They will not be quietly reprimanded at a review in a year. The entire organisation’s direction is what is at stake. The way the project was announced, the framing of it as something handed off so the leader could go back to business as usual, was itself the first sign it had failed. That is what I held back that day. The honest alternative is to decide, before it starts, that you will own the discomfort and protect the people doing the real work.
If you want help owning an AI initiative properly rather than delegating it into failure, that is the conversation I have with leaders.
Frequently asked questions
Is delegating an AI project to a capable person the right move? Not if the leader then steps back. Leading an AI project is not something you hand to a likeable person and admire from a distance. The most common failure is exactly that: the work is handed off, announced as a win, and the leader returns to business as usual while the company’s future rides on a project they have stopped touching.
Why is choosing the most fluent person to lead AI a warning sign? Because choosing fluency over substance suggests the leader cannot tell the difference, which usually means they have not done the work themselves. An AI project is not a communications exercise; it is an innovation effort that has to change how the company works. The person who can drive that is often not the most comfortable to be around, because they tell truths people do not want to hear.
What is the KPI question that exposes a failing AI project? Ask the leader directly what their own KPIs are, and what standard they are holding themselves to while the project runs. The measure that matters is customers, whether they are more profitable, more efficient, given genuinely new models. If the real measure is whether the leader avoided embarrassing themselves, the project has already failed.
How involved does a leader actually need to be? Every single day, not running the project but in it, side by side. It is the leader’s future on the line, not the project owner’s, and the entire organisation’s direction is what is at stake. The owner will not be quietly reprimanded at a review in a year; the whole company will feel the outcome.
What is the difference between delegation and abdication here? Delegation shares the work while the leader stays accountable and present. Abdication hands the project off so the leader can go back to business as usual. Framing an AI initiative as something handed away is itself the first sign it has failed, because the leader has separated themselves from the future that depends on it.
Thomas Anglero is a Strategic AI Advisor, keynote speaker and author of Intro to Artificial Intelligence. He has delivered over 450 keynotes across 30 countries for organisations including IBM, the WHO, the World Government Summit and the European Commission. He founded the IBM Watson AI Lab for Cancer at the Oslo Cancer Cluster and closed over $500 million in enterprise transformation deals as CTO and Chief Innovation Officer at Cognizant.