There's a concept circulating right now called "AI psychosis." Box CEO, Aaron Levie put a name to something a lot of people have been sensing: leaders who demo AI tools, see the happy path, and leap straight to believing agents can handle everything. They skip over the last mile entirely; the review, the debugging, the edge cases, the human judgment that still has to happen before any output delivers real value.
It's a true phenomenon, but... it's only half the story.
Because while some leaders are overconfident about AI, there's a quieter problem running alongside it — and it's one that gets a lot less airtime. Leaders who are effectively disconnected from how AI is actually functioning across their teams. Big decisions are made by leadership who don't have a clear-eyed view of what the operational reality genuinely looks like.
Spoiler: 54% of marketers think their leadership underestimates the human effort AI actually requires. More on that in just a scroll.
We're calling it AI distance, and it's happening a whole lot more than we're comfortable to admit.
"AI? I'm too old for that"
It's a line that keeps coming up. Not shouted, not defensive — said almost as a statement of fact, by people who are smart, senior, and fully bought into AI as a strategic priority. The conversation turns to actually picking up a tool and solving something specific, and they step back with the defeatist "I'm too old for that" or "I don't think I'll do it right".
The leaders who step back and leave it to the (sometimes young, sometimes not, but always enthusiastic-to-learn) practitioners reveal even more about the gap we're talking about.
And it tracks, when you think about where the appeal actually sits. For a practitioner, the case for AI is right there in front of them — less hours lost to first drafts, less of the boring admin stuff, more time for the work that actually feels like work. The benefit of agents in their day-to-day is immediate and obvious, which is exactly why they want to play around with it.
For many leaders, that case is way less visible. AI can't sit in a room and mediate between two department heads who disagree. It can't read a face-to-face conversation and know when someone needs reassurance instead of a status update. It can't stand in front of a board and make the call. So much of what senior leadership actually does all day is exactly the work that AI hasn't touched — which makes "why should I bother" feel like a perfectly reasonable response for some.
But here's the part that gets missed: there's a second benefit to hands-on experimentation that has nothing to do with personal productivity. It's understanding. Knowing — not being told, but knowing — what your team is actually living through. The friction, the prompting trial-and-error, the moments AI gets it embarrassingly wrong. The knowledge doesn't transfer secondhand. And without it, leadership ends up setting direction for a reality they've never actually stood inside.
It's a gap that's quietly opening up inside so many marketing organizations right now. The people setting direction for AI adoption, approving the tools, championing the strategy are not the ones who have to live inside those tools day-in, day-out. And that distance matters more than it might seem.
Using AI and understanding AI aren't the same thing
Anyone who's been close to an AI rollout in the last couple of years (you, you, and.... you?) will recognize this pattern:
Move #1: Organizations start by throwing open the doors. Hackathons, daily experimentation, everyone encouraged to build their own agent.
Result: It's energizing, curious, creative... and completely ungovernable. No one really know what's working, there's no consistency — but plenty of chaos, if you're into that.
Move #2: So, the pendulum swings the other way. All the way towards centralized ownership, standardization, governance frameworks, tighter control.
Result: That momentum you were building? Gone. Deceased. RIP. 🪦
The organizations that are getting AI right are landing somewhere in the middle. This is the sweet spot where teams are still building, still experimenting, but within a structured and orchestrated system so things don't get too wild. We're sitting our butts right down, in between maximum freedom and maximum control, on a chair of informed autonomy.
The same dynamic plays out at the individual leadership level. The 'AI psychosis' end of the spectrum is one kind of failure, but the retreat — AKA the opting out, the "I'll leave that to my team," the learned helplessness dressed up as delegation — is another.
Both create the same fundamental problem: a leader whose mental model of what AI can and can't do is disconnected from reality. And there's proof in that. 👇
It's a real nice view from up there
We surveyed 2,003 B2B marketing leaders across seven global markets, and the gap between what leadership thinks is happening and what's actually happening on the ground is, frankly, a very big gap indeed.
- 54% say their leadership underestimates the human effort AI actually requires
- 76% of practitioners spend 3+ hours every single week fixing, fact-checking, and editing
- 4% said AI genuinely saves time at every stage of the marketing lifecycle — just 4%
Meanwhile, 49% of C-level respondents rate their own expectations as "highly realistic". Might be time to reconsider that one.
And folks, this is NOT a communication problem... it's a proximity issue.
When leaders aren't close enough to the work to feel the friction — the hallucination review, the constant copy-pasting between tools that don't talk to each other, the deadline pressure that leads 25% of marketers to publish AI content they know isn't on-brand (say WHAT?) — things look a whole lot more optimistic. Optimistic in a way that's genuinely expensive for the people doing the work underneath them.
The view really is nicer from up there. It's just not necessarily the view that's true.
Understand the whole story with our recent global study, showing the gap between what leaders are thinking versus what's really happening.
So, what does 'close enough' actually look like?
Disclaimer: we are not pitching that every VP needs to become a power user or start vibe-coding their own agents in their spare time. AI-close leadership isn't about becoming a practitioner. It's about being informed enough to lead it well.
That means having a honest view of where the real friction in your team's work actually lives; not the version that gets smoothed over by the time it reaches you. It means being able to look at an AI use case and ask the right question. Not "what could AI do here?" (a question that tends to produce answers that are either overwhelming, superficial, or wishlist-like) but "what does this workflow actually look like end-to-end, and where does the value really come from?"
That's a question that requires leaders to be close enough to the work to answer it honestly. Which is exactly why we built a structured way to get there.
Start with the use case, not the tool itself
If any of this is landing a little close to home, the best next move isn't downloading another tool. It's getting specific about where value actually lives within your team's workflows.
The AI Use Case Discovery Template is a practical starting point: a structured way to identify where AI can make the most meaningful difference for your specific team, based on what you're actually trying to achieve. Built for marketing leaders who want to move from "we really should be doing more AI" to a concrete, prioritized view of where to focus.
Because the question was never whether AI can do something — we already know that very well. The question is whether you're close enough to the work to know what's actually worth doing.