AI & Digital Change · The leadership behind transformation
AI adoption is a leadership challenge
By Isabella Brusati
Making AI available is a technology decision. Making it useful in everyday work also requires leadership decisions about purpose, judgment and responsibility.
For a CEO, CIO or COO, the question is not simply whether employees are using a tool. It is whether the organisation understands where it should be used, what good use looks like and how people will remain accountable for the work.
Without that clarity, a rollout can generate activity while leaving fundamental questions unresolved. Enthusiasm and hesitation may coexist for entirely sensible reasons.
Start with a specific piece of work.
“We need to adopt AI” is too broad to guide a team’s daily decisions. Begin with a defined task or workflow and a clear explanation of the problem you want to address.
For example, a team might explore whether an approved tool can help produce a first draft of an internal briefing. That is a different proposition from allowing a tool to make a consequential decision or communicate directly with a client. The required review, controls and responsibilities will differ.
Agree what improvement would mean before starting. It might involve reducing repetitive preparation, improving consistency or creating more time for analysis. Include quality and review effort in the assessment: a faster first draft is not necessarily a better overall process.
Make human responsibility explicit.
People need to know who checks an output, who can approve its use and what happens when the result is uncertain or unsuitable. They also need clear organisational guidance about approved tools and appropriate information.
These decisions should involve the relevant operational, technology, security and governance specialists. A general encouragement to “experiment” is not a substitute for boundaries that people can understand and apply.
Managers have a particular role. They must be able to discuss the quality of the work, make room for learning and reinforce the expectation that outputs are questioned rather than accepted automatically. Responsibility becomes harder to exercise when the organisation celebrates speed but leaves the cost of review invisible.
Understand what adoption asks people to reconsider.
Introducing AI can change more than a task. It may raise questions about expertise, professional identity and the future shape of a role. A person who has built credibility through specialist knowledge may reasonably ask how that contribution will now be recognised.
Leaders do not need to promise certainty they do not have. They do need to explain what is known, what remains undecided and how people will be involved as the work evolves.
Listen for practical concerns beneath broad labels such as fear or reluctance. Is someone unsure how to assess an output? Worried about accountability? Unable to find time to practise? These concerns call for different responses.
Build learning into the work.
A launch session can introduce a tool. It cannot anticipate every judgment people will need to make when using it.
Create opportunities to work through realistic tasks, compare results and discuss limitations. Encourage people to bring examples that did not work as well as those that did. If only success stories are welcomed, leaders may receive an incomplete picture of where the approach is useful.
Keep the learning focused. A bounded trial with clear responsibilities can provide more useful evidence than a broad rollout whose purpose is unclear. Decide in advance when to continue, adapt or stop, and who will make that decision.
Measure what matters after the launch.
Usage can help show whether people are engaging, but it does not establish value on its own. Look at the task end to end: preparation, review, correction, handover and the quality of the final result.
Ask managers what has changed in practice. Are people using their judgment more effectively? Have new bottlenecks appeared? Is the organisation learning where the tool is unsuitable as well as where it helps?
These are questions of organisational adoption. They connect the technology investment to how work is actually performed and governed.
The executive move
Before expanding your next AI initiative, ask its owner to explain five things in plain language: the task being improved; the intended benefit; the human decision owner; the support and boundaries people need; and the evidence that will determine the next step.
If the answers are unclear, resolve them before treating wider access as progress.
A question to take away: Are you asking people to adopt a tool, or helping them understand how to do valuable work differently?
An original leadership perspective on AI adoption. The briefing example is hypothetical. These practical questions complement the organisation’s specialist technology, security and governance advice.