When boards talk to leadership about AI, the conversation usually stops at where AI could be useful. It needs to go two steps further: a realistic timeline for implementation, and a plan for the change management and behavior change that comes with any real technology shift. Boards that skip those last two pieces end up approving strategies that look complete on a slide and stall the moment they meet actual employees doing actual work.
Boards are talking about AI. They’re not equipped to talk about it well
AI now comes up in the vast majority of board and management meetings. That’s not really in question anymore. What’s less certain is whether those conversations are useful. In a McKinsey survey of corporate directors, 66% said their boards have limited to no knowledge or experience with AI, and nearly one in three said AI doesn’t even show up on their board’s agenda.
That’s the gap Merav Yuravlivker, Chief Learning Officer and Co-Founder of Data Society, sees showing up in how boards actually frame these conversations with leadership:
“As the board is speaking to leaders around AI use, it’s important to have it lined up in terms of where AI can be most impactful in a positive manner, and what the timeline is for that implementation, and then that change management and that behavior change aspect that really comes with the territory of these technological shifts.”
Merav Yuravlivker, Chief Learning Officer and Co-Founder, Data Society
Notice the structure in that statement: it’s three distinct pieces, not one. Impact. Timeline. Change management. Most board conversations about AI get through the first one and stop.
Part one: impact, done specifically
“Where AI can be most impactful” sounds obvious, but done well it’s a narrowing exercise, not a brainstorm. The useful version of this conversation isn’t a list of every process AI could theoretically touch. It’s a short list of where AI will move a number the business actually tracks: cost per outcome in a support workflow, cycle time in a review process, error rate in a compliance check. Boards that skip the narrowing step tend to approve broad, vague AI mandates that sound ambitious and produce nothing measurable a year later.
Part two: a timeline leadership will actually hold to
The second piece, a real implementation timeline, is where a lot of AI strategy quietly falls apart. It’s easy to greenlight “we’re going to use AI in customer support.” It’s much harder to commit to a date range, name the milestones in between, and hold the organization to them. Without that specificity, “AI implementation” becomes an open-ended initiative that never quite finishes and never quite fails. It just fades, matching the quiet-failure pattern that shows up whenever nobody is tracking whether a deployed system is actually being used.
Part three: the piece boards skip most
Change management is the part Yuravlivker’s framing puts last, and it’s also the part boards are least equipped to discuss, precisely because it isn’t a technology question at all. Deploying a new model or tool is a project. Getting people to actually change how they work, trust a new system, and give up old habits, is a much longer and messier process, and it’s the piece that determines whether the first two parts of the plan produce anything.
The data backs this up directly: boards themselves report that AI literacy and cultural resistance, not the underlying technology, are the real constraints on getting AI initiatives to land. A board that can speak fluently about model capabilities but has no answer for “how do we get our people to actually adopt this” hasn’t finished the conversation. It’s stopped at the easy two-thirds.
Why all three have to be in the same conversation
These three pieces aren’t independent line items. They shape each other. A narrow, specific impact target makes the timeline realistic, because you’re not trying to change everything at once. A realistic timeline makes change management possible, because people need time to adjust, not a mandate that lands all at once. And a real change management plan is what makes the impact target achievable in practice rather than just on paper.
Leave any one of the three out, and the other two don’t hold up. An ambitious impact target with no change management plan produces a tool nobody uses. A careful change management plan with no clear timeline produces an initiative that drifts indefinitely. A tight timeline with a vague impact target produces a rushed rollout of something that was never well-defined to begin with.
What this looks like in a real board conversation
In practice, a board-ready AI conversation should be able to answer three questions in sequence, not just the first one: Where specifically will this AI initiative move a number we track, and how will we know if it worked? What is the actual timeline, with named milestones, not an open-ended “we’re working on it”? And what is the plan for getting the people affected by this change to actually adopt it, including who owns that work and how success will be measured?
If a board can only answer the first question, the strategy isn’t ready for leadership yet. It’s still a wish list.
Frequently asked questions
What should a board’s AI strategy conversation with leadership include?
Three elements: a specific, measurable area of positive impact, a realistic implementation timeline with milestones, and a change management plan for how employees will actually adopt the new way of working.
Why is board AI expertise a concern right now?
In that same McKinsey survey, 66% of directors said their boards have limited to no knowledge or experience with AI, making it difficult for boards to push back effectively on vague AI proposals.
Why does change management matter more than the technology itself?
Boards and researchers increasingly point to cultural resistance and AI literacy gaps, not technical limitations, as the real barrier to AI initiatives succeeding. A technically sound deployment still fails if the people using it don’t adopt it.
What happens when a board skips the timeline or change management pieces?
Initiatives without a specific timeline tend to drift indefinitely without ever being declared a success or failure. Initiatives without a change management plan often see the technology deployed but not actually used, quietly wasting the investment.
How can a board tell if an AI proposal is actually ready for approval?
It should be able to name a specific, measurable impact target, a realistic timeline with milestones, and a concrete change management plan with clear ownership. If it can only speak to the first of these, it isn’t ready yet.
Next steps
Boards have gotten comfortable asking where AI could be useful. That’s the easiest third of the conversation. Data Society’s AI Advisory work helps boards and leadership build the timeline and change management pieces most conversations skip, and AI upskilling programs build the AI literacy boards need to ask sharper questions.
