Who owns AI at your organisation? Is it the CIO, who frames it as an infrastructure project? The Chief People Officer, who treats it as a change programme? The newly appointed Chief AI Officer, who is still learning how your organisation works and how to get things done in it: influence without authority, mandate without context? Or has nobody been given the role, and AI has become everyone's side project: a dozen tools, no policy, no shared learning?
The question matters. AI will be one of the most consequential, multi-faceted transformations most of us go through in our working lives. It requires coordination across the organisation and governance at the technology, financial, and people layers. Globally, CEOs have recognised this. BCG's latest survey of 2,360 executives found that nearly three-quarters of CEOs now say they are their company's chief AI decision-maker, double the share from a year ago. That survey skews toward large enterprises across 16 markets. Data on smaller firms is harder to find, but the Reserve Bank of Australia's assessment last November found that where AI adoption is happening in smaller firms, it is often employee-led rather than employer-led. Staff are experimenting. Leadership isn't directing.
Whatever the size and shape of your organisation, the question of which leader should play what role is beside the point if you can't answer two fundamental questions first.
What is AI for at your company?
Not the miracle that will 10x your team, and not the buzzword that keeps the board off your back. What role does AI actually play in your business, and what role do you want it to play?
There are really only three answers.
AI reduces your cost of delivery: same services, produced faster and more consistently. AI changes what you can offer: faster turnaround, broader coverage, service formats that weren't viable before. Or AI changes how you select and serve clients: same value proposition, different competitive position because you can take on work you previously couldn't staff profitably.
Most leaders I talk to describe the second but fund the first. That gap is where credibility erodes. A team that hears "AI will transform what we do" but only sees investment in automating routine work draws its own conclusions about what leadership actually means.
Every leader I've put this question to has an immediate answer. When I ask what they've funded in the last twelve months, the answer usually changes. The gap isn't just analytical. It's experiential. A leader who hasn't worked with AI on their own strategic problems is unlikely to know where AI is a useful accelerator, a powerful augmentor, or an unreliable hallucinator of facts. Without that direct experience, they don't have the reference points to judge what AI does well and where it falls apart in their specific context. Spencer Stuart's January 2026 playbook for CEOs puts it directly: "The CEOs who are making better, faster decisions with AI are power users — not delegators."
There's a subtler risk, too. Leaders with moderate AI exposure hit a peak of confidence that deeper experience corrects, a pattern researchers have linked to the Dunning-Kruger effect. They've seen enough to form a view but not enough to know where it breaks down. Claiming you've chosen transformation while funding efficiency isn't always a deliberate gap. Sometimes the leader genuinely believes their own framing. They haven't tested it against real use, don't have the measurement and feedback loops in place, and have hired a workforce to enable an ambition they haven't validated. But that's another article.
What do you say when someone asks how you use AI?
It might be a sophisticated customer, a board member asking about risk, or a prospective hire wondering whether the company is serious or just paying lip service. Someone is going to ask, probably sooner than you think.
Could you answer in thirty seconds, without jargon, and have the person nod rather than ask a follow-up?
There are three honest positions. Proactive: you include your AI practices in engagement terms or board reporting as standard. On-request: you answer honestly when asked but don't volunteer it. Context-specific: you disclose for certain work but not others. Each position is defensible. Not having a position is not.
This question is different from the first one, and harder than it looks. The first question can stay internal. It can evolve quietly as the organisation learns. This one can't. The moment you state a position to a client, a board, or a regulator, you've made a commitment you can be held to.
And the environment is forcing the question whether firms are ready or not. Customers are asking, particularly sophisticated institutional and government buyers who are developing their own AI positions. Regulators are tightening. In Australia, new Privacy Act transparency obligations for automated decision-making start in December 2026. Prospective hires are using your answer as a filter for whether the organisation is genuinely serious about AI or just talking about it. A vague answer in any of these conversations isn't neutral. It signals that the company hasn't done the thinking.
The question also forces you to resolve things that most businesses haven't. A law firm that uses AI to draft and a human to verify might ask: what are we actually charging for? If a regulator asks about your AI practices and you don't have a documented answer, what does that say? If you disclose to one client and not another, can you sustain that?
Unlike the first question, which can evolve privately, this one is on the record. And once the answer is on the record, it constrains what follows.
The questions come before the org chart
This is the pattern underneath the ownership debate. A CEO recognises AI matters. They appoint someone. The person they appoint inherits a mandate shaped by a mental model the CEO formed without enough direct experience to form it well.
If the implicit model is that AI is a cost-reduction tool, the CAIO builds a cost-reduction programme. If the model is that AI is a productivity play, the rollout focuses on efficiency metrics. The delegation doesn't correct the frame. It encodes it. BCG's data shows the 15 per cent of CEOs driving real AI results spend more than eight hours a week on their own AI upskilling and invest twice as much as their counterparts in capability-building across their organisations. They are not delegating and walking away. They are building the judgment to set a direction worth following.
The ownership question will need an answer. But the leaders who answer it well will be the ones who worked with AI enough to know what they're asking someone else to do.
Start there. Work with AI on a real strategic question, something that matters to your business, not a sandbox exercise. The two questions above will get easier to answer, and the ownership decision will follow from judgment, not from the org chart.
Sources
BCG, AI Radar 2026: As AI Investments Surge, CEOs Take the Lead (January 2026). 2,360 executives across 16 markets and nine industries, including 640 CEOs. https://www.bcg.com/publications/2026/as-ai-investments-surge-ceos-take-the-lead
Reserve Bank of Australia, Technology Investment and AI: What Are Firms Telling Us? Bulletin, November 2025. https://www.rba.gov.au/publications/bulletin/2025/nov/technology-investment-and-ai-what-are-firms-telling-us.html
Spencer Stuart, Don't Delegate AI: A Power-User Playbook for CEOs (January 2026). Jason Baumgarten, Amanda Morgan McAllister, and Kim Spalding. https://www.spencerstuart.com/research-and-insight/dont-delegate-ai-a-power-user-playbook-for-ceos
Guan, J., He, X., Su, Y. and Zhang, X-a. (2025), "The Dunning-Kruger effect and artificial intelligence: knowledge, self-efficacy and acceptance," Management Decision, Vol. 63 No. 10, pp. 3786-3802. DOI: 10.1108/MD-06-2023-0893. https://www.emerald.com/md/article-abstract/63/10/3786/1259025/The-Dunning-Kruger-effect-and-artificial
Privacy and Other Legislation Amendment Act 2024 (Cth). Automated decision-making transparency obligations (APP 1.7-1.9), commencing 10 December 2026. https://www.oaic.gov.au/privacy/australian-privacy-principles/australian-privacy-principles-guidelines/chapter-1-app-1-open-and-transparent-management-of-personal-information