You’re in a board meeting or an investor update, and the conversation turns to AI. Someone mentions agentic workflows or multi-model orchestration. You nod, because that’s what you do. You’ve been nodding since the GPT-4 announcement, and by now the nod arrives before you’ve fully processed the question.
That nod is a decision. It tells your delegate that you won’t push back on their proposal. It tells the vendor that the room is receptive. It tells your board that you’re across things. None of it is quite true, and all of it shapes what happens next.
The fear of appearing out of touch on AI is doing more damage than actually being out of touch.
What does “looking out of touch on AI” actually mean for a founder?
For a founder in an investor-backed business, AI literacy has become a proxy for being in command. Investors and board members carry a version of the expectation that the CEO should understand this. Admitting you don’t, even privately to a trusted COO, can feel like handing someone a reason to question your judgement precisely when AI dominates every board deck.
That pressure is not imaginary. Founders who built their businesses on domain expertise feel particularly exposed when they step into territory where their own reports know more. In an investor-backed context, the exposure is sharpened by the dynamics of the room: you answer to a board, you carry a reputation, and appearing forward-facing is part of the deal you made when you took the money.
BCG’s research on board attitudes toward AI captures the dynamic directly. Boards and CEOs broadly agree that AI should sit with the executive team, yet many boards implicitly expect the CIO or CTO to own execution. What emerges is a performance of strategic engagement without the substance behind it. A founder who looks switched on. A board that looks rigorous. Nobody actually making the call.
Research on founder psychology has consistently found that status exposure of this kind, feeling like a novice in front of people who report to you, activates avoidance rather than inquiry. The impulse is to manage the gap rather than close it. And managing a gap you haven’t admitted produces a very particular set of decisions.
Why does this status anxiety make your AI decisions worse?
When you can’t admit a gap, three things happen that directly corrupt your decision-making. You delegate to look modern rather than because the delegate is set up to succeed. You approve proposals you haven’t fully grasped. You defer to vendors who read your uncertainty and fill it with confidence. Each decision looks reasonable on its own. Together, they leave your AI programme without an owner.
The failure rate for generative AI pilots is relevant context here. MIT’s NANDA group found roughly 95% of such pilots stall or show no measurable P&L impact. The common failure point is leadership and workflow integration, not model quality. A founder who stays at arm’s length from their own AI programme is precisely the sponsorship gap the research predicts.
The delegation problem also has a secondary layer. When you hand the AI mandate to a COO while signalling, through your own behaviour, that you won’t look closely, you remove the strongest lever available. Sustained, visible sponsorship from the top is what change management research consistently identifies as the difference between adoption that sticks and adoption that stalls. Without it, the AI programme drifts between your delegate’s ambitions and the team’s actual habits.
Where does the fear show up in practice?
The clearest sign is the vendor meeting where you don’t ask the obvious question. A confident pitch lands, the numbers look right, and rather than expose your uncertainty, you move to next steps. Your delegate’s follow-up confirms the direction you’ve already signalled you want. The decision is made before anyone with full information has had the chance to make it.
A second sign is the approval that happens by default. The proposal lands in your inbox. You read the executive summary, skip the technical section you weren’t sure how to evaluate, and approve the spend. Three months later the implementation is struggling, and the post-mortem conversations are uncomfortable, because the person who signed off the budget didn’t fully understand what they were buying.
Conor Grennan, who has worked extensively on AI literacy for senior leaders, points to a root cause that makes this pattern harder to break than it looks. Treating conversational AI “like software to be learned” is a category error. Excel and PowerPoint are procedural. You learn them through drills. Using AI well requires a shift in how you work with information, how you frame questions, how you evaluate outputs. That shift happens through use, not through watching someone else use it. A founder who delegates the learning away with the programme never gets the shift, and never gets the judgement that comes with it.
When does admitting the gap become the smarter call?
The risk of admitting uncertainty is real, but smaller than it feels. BCG found roughly 60% of CEOs feel their boards are rushing AI, and those pushing hardest are often the least confident, overcompensating for their own gaps. Saying “I need to understand this properly before I approve it” is a reasonable position to hold. Approving without understanding is the riskier move.
The right moment to admit the gap depends less on the audience and more on what you’re being asked to decide. Budget approval for a vendor you haven’t properly assessed is worth slowing down. A board update where you’re asked about your AI roadmap is worth an honest answer about where you actually are. The founders who come unstuck are the ones who treat both situations as performance rather than decision-making.
Kyndryl’s 2024 research found around 70% of senior leaders say their workforces aren’t ready for AI, yet only 14% have aligned their workforce planning, technology choices, and growth objectives. The alignment failure often starts at the top, with a leader who is managing appearances rather than building understanding. A leader who genuinely grasps what they’ve approved makes better calls downstream. The change management evidence supports that directly.
What does safe, private learning actually look like?
Spencer Stuart’s research on CEO AI capability points to a direct entry position. Before any company-wide push, start with one task you already do yourself. A pre-meeting summary, a daily brief, or a first draft of a proposal. Something private, low-stakes, with no audience watching. That personal experiment does more for your credibility as a decision-maker than any amount of alignment conversations with your AI lead.
There is also a practical change in how you show up to the conversations that follow. There is a meaningful difference between “I haven’t really used it myself” and “I’ve been working with it on a few of my own tasks and here’s what I’ve noticed.” The first confirms the gap the fear was protecting. The second positions you as someone in the process of learning, which is both true and considerably more credible with the people around you.
Research on perceived control and wellbeing is relevant here too. Perceived loss of control is identified as one of the more destabilising experiences for leaders. Using AI personally, even in small ways, returns direct understanding that makes the wider programme feel less like something happening around you and more like something you’re actually directing. That is not a minor psychological benefit. It changes the quality of the decisions you’re able to make about it.
The fear is real. Boards expect AI competence. Investors ask about it. Your senior team assumes you know more than you’ve said. But the response to that pressure, the nodding, the approving, the deferring, is costing you the thing you actually need.
The founders who get this right are not necessarily more technical. They’re the ones who found a private route into genuine understanding before anyone asked them to go public with it. That’s a much cheaper path than the one the bluff produces.
If you’d like to think through what that looks like for your specific situation, book a conversation.
