Who Actually Leads Your Company's AI Strategy?

63% of CEOs think their board overestimates what AI can actually do. That statistic comes from BCG's first survey pairing CEOs and board members from the same companies, 625 leaders in all, published in April 2026. I want to sit with that number for a second. It means most CEOs believe the people overseeing them are working from a different set of facts.

Then the survey asks who actually leads AI implementation. 47% of CEOs say they do. Only 39% of board members agree. Same companies, same question. The two groups can't settle who's driving. I've sat across tables where a board was certain it understood a technology shift. The executives across from them were just as certain the board was three steps behind. BCG measured that exact pattern at scale for the first time. The gap is wider than most rooms admit out loud.

Picture the actual meeting behind these numbers. A board member asks a pointed question about the AI rollout timeline. The CEO answers with confidence. She assumes the board already signed off on the plan months ago. The board member nods. He's not sure whether to push further or just trust that someone else already vetted it. Nobody in that meeting feels behind. Everyone leaves believing someone else owns the next decision.

Here's the part that surprised me most. The boards feel behind, not the CEOs. 75% of board members rate their own AI literacy as sufficient or advanced. Among the ones who admit they're less AI-savvy, 80% still worry the company isn't moving fast enough. The people with the least confidence in their own knowledge push hardest for speed. Anxiety is setting that pace, not caution.

I ran a $1.1B innovation portfolio across 14 countries at Cisco. That put me in front of more boardrooms than I can count. The boards that helped a transformation succeed were rarely the ones that understood every technical detail. They agreed, in the room, on two plain things. Who owns the decision. And how we'll know it worked. BCG's data shows most companies still haven't settled either one for AI. 35% of CEOs weigh AI ROI in their own performance reviews. Only 27% of boards hold themselves to that same bar. A strategy nobody gets graded on stays a slogan, no matter how often it's repeated in the deck.

79% of CEOs and board members support requiring AI literacy for future board candidates. Almost everyone in the survey agrees on the fix. Almost nobody has put it in place yet. That gap between what leaders know they should do and what they've actually done shows up everywhere I look in enterprise AI right now, not just in this one survey.

I think about this the way I think about every major shift boards have had to absorb before. Most board members weren't accountants either when companies first added financial risk committees in the 1990s. The fix wasn't sending every director back to school. Someone owned the risk conversation, and that person answered for the results. AI governance is running through the same lesson now, just faster and more publicly than risk committees ever did.

The relevance cliff doesn't wait for a boardroom to agree on who's in charge. Skills expire on their own schedule, whether or not the CEO and the board have settled their disagreement about pace. Every quarter this stays unresolved is a quarter where somebody in the company makes real AI decisions without a clear mandate, while everyone above them assumes someone else already gave it.

Boards spend real money finding out who was actually accountable only after a decision goes wrong in public. A smaller, quieter fix costs nothing and works before that happens. Put one question on your next board agenda instead of a slide deck. Who owns this quarter's AI decisions, and what result will we check for next quarter? If the CEO and the board answer differently out loud, in the room, you've found the real problem before it costs you a year.

Most boardrooms don't learn they're divided until a decision fails where customers or regulators can see it. Which way would your last three AI decisions have gone if you'd asked that question first? If you're not sure, that uncertainty is the actual risk here, not the technology underneath it.

Sources: BCG, "Split Decisions: The BCG CEOs and Boards Survey," April 2026 (625 leaders, 351 CEOs and 274 board members, companies with $100M+ revenue).