Only about 14% of senior leaders sit in the top third of AI tool deployment, according to leadership consultant Dusty Holcomb — and the rest, he says, are "lagging" with no clear starting point. His prescription for closing that gap is organizational, not technical: put HR in charge of the rollout, not the policy file.

Holcomb, a US-based former chief executive who has just signed a book deal on leadership in the AI era, and Julie Coleman, consultant for people and culture at Australian battery recycler Livium, told HRD that AI adoption succeeds when companies treat it as a leadership and change-management problem. HR teams, they argue, are best positioned to run the process — but only if they use the tools themselves rather than administer rules about them.

What does leading, rather than supporting, look like?

Holcomb said the most common executive mistake is separating AI from people leadership. "You need to stop thinking about leaders, tools and artificial intelligence," he said. "You need to start thinking about how you're going to lead people, because if you lead people well, then you will be able to lead the machines and the tools well, because the principles are exactly the same."

Coleman's experience at Livium shows the approach in practice. The trigger for formal AI governance came in mid-2025, when employees had already started using generative tools such as ChatGPT on their own initiative. Rather than respond with a blanket policy, the company ran a structured pilot with a small group of users, tracking time saved on tasks and error reduction over several months before widening access. Livium now reports on AI use to its board every quarter.

How was governance built?

Livium anchored its framework in the Australian Institute of Company Directors' eight elements of safe and responsible AI governance, developed with the University of Technology Sydney's Human Technology Institute.

"We went back to the AICD principles … just to again make sure that from a senior management and board perspective, that we were putting the right infrastructure and governance in place when implementing it," Coleman said.

That meant revisiting existing privacy, cybersecurity, discrimination and work health and safety policies to confirm they covered generative AI, instead of drafting standalone AI rules. The connective role has since extended into cybersecurity: Livium has folded AI use into its incident-response and cyber-insurance arrangements as a proactive measure.

Training relied on a simple framing that Coleman said resonated with staff: think of AI as "a brilliant graduate on day one" — capable, but needing careful prompting, fact-checking and direction rather than blind trust.

"It's not rocket science, to be frank," she said. "But it's just making sure that it really does integrate with what you already have in place."

Why do usage mandates backfire?

Holcomb was blunt on the alternative. "Mandates that people use the tools are an exercise in misguided futility," he said, warning that poorly designed usage targets produce activity without outcomes.

The 14% figure comes from his own conversations with chief executives and reflects a market split between a small advanced cohort and a majority unsure where to begin. The implication for HR leaders: adoption stalled at the mandate stage is a change-management failure, not a technology one.

What is HR's actual mandate here?

Both leaders converged on the same conclusion. HR's job is not to police AI use but to translate it into outcomes employees understand.

"HR is kind of the connective tissue," Holcomb said, arguing that HR leaders need to use the tools first so they can "ask better questions of the teams" and address the fear that often accompanies automation of routine work.

For HR functions weighing their next move, the Livium model suggests a sequence: pilot with a small user group, measure time and error outcomes, embed governance in existing policies, report to the board — and have HR practitioners build their own fluency before scaling access. As more organizations formalize AI use over the coming year, the ones that stick, these leaders suggest, will be those where HR led from the front rather than filed the policy.