The Imagination Ceiling: Why 70% of Your People Are Stuck on AI

The Imagination Ceiling: Why 70% of Your People Are Stuck on AI

I analyzed 5,000 anonymous survey responses looking for the AI adoption blocker. I found it. And it's not what I expected.

For the past 18 months, I've been surveying people before they enter my 17-day personal AI bootcamp course. Simple question: "What's your biggest barrier to using AI more effectively?"

The answers cluster into four buckets:

"I know it could be helpful, but I don't know HOW to apply it to my job" — ~70%

"I just forget to use it" — ~10%

"I'm nervous about doing something wrong" — ~10%

"I don't think AI can help me" — <5%

Look at that last number. The skeptics are basically gone. Only 4% don't believe AI can help them. Meanwhile, 78% describe themselves as "excited" or "cautiously optimistic."

Your people aren't skeptics. They're believers who are stuck.

What I found was a ceiling—not on capability, but on imagination.

And here's what's been nagging at me: the "how" problem is getting worse, not better.

Q4 2024: 75% of respondents didn't know HOW

January 2026: 83% of respondents didn't know HOW

Despite 18 months of tutorials, courses, podcasts, and corporate AI initiatives—we're going backward. (And yes, that trend line surprised me too.)

The Paradox Nobody's Talking About

I assumed the "how" problem would shrink with usage. Use AI more, figure out how it helps your job, problem solved. Right?

Except even among people who use AI many times per day, 64% still don't know how to apply it effectively. Even among people who rate AI's value at 4 or 5 out of 5—believers!—68% are stuck on the "how."

Using AI more doesn't teach you how to use AI better.

And as I've said before, more use isn't better; better use is better. (I know—that's uncomfortable if you've been measuring "adoption" by login frequency.)

There's a ceiling at work here. And it's not where most people think.

What the "How" Problem Actually Looks Like

I dug into a second dataset—1,376 responses from the AIQ self-assessment—to understand what "not knowing how" looks like in practice.

The hardest behaviors aren't technical. They're relational.

Using AI as a sparring partner—someone to debate with, push back on, challenge and be challenged by? 51% never do it. Only 8% do it often.

Using AI as a sounding board to validate ideas? 35% never. Only 15% often.

Brainstorming and generating creative content? 24% never. 22% often.

Look at that first line again. Only 8% regularly use AI as a sparring partner. Eight percent.

The Iteration Gap

There's another pattern buried in the data that might matter more than anything else.

More than half of all respondents stop after just 1-2 prompts. Only about 10% go beyond five.

Here's why that matters: People who use 5+ prompts save 5+ hours per week. People who stop at 1-2? Less than an hour.

Do the math. That's the difference between a search and a collaboration. That's the difference between where you are now and having six extra weeks a year. (Real talk? If AI isn't saving you 5+ hours a week, you're probably treating it like a search engine.)

The people who've cracked this have learned to treat AI more like a capable and tireless junior team member. When they approach AI co-working as a management undertaking rather than a matter of tool use, the number of turns in the conversation increases dramatically.

You can't imagine a sparring partner if you've only ever seen a search bar.

The Ceiling

I know "imagination" sounds soft. Bear with me.

In my surveys, I ask people where they get ideas and inspiration for using AI—podcasts, YouTube, newsletters, colleagues, whatever.

Nearly 40% select "none of the above."

No sources. No one showing them what's possible. Flying completely blind.

Why? Microsoft's research offers a clue: people who use AI are often perceived as "lazier, less competent, less diligent, less trustworthy." Being seen using AI can actually harm your relationships with colleagues.

So people don't share what they're doing. The edges stay quiet. Best practices stay invisible.

I was talking to a head of L&D at a Fortune 500 last month, and she said something that stuck with me: "Our best AI users are in the closet. They're afraid of being seen as cheating."

That's the ceiling. Most people have only ever seen AI used like a search engine—one query, take the first result, move on. That's their entire mental model.

Why Training Isn't Working

This explains why most AI training programs are failing at scale. (Sorry if you just bought one.)

We're teaching prompting tips. We're teaching features. We're teaching the wrong thing entirely.

You can't information-transfer your way to imagination. You can't teach someone to see AI as a sparring partner through a slide deck. They have to witness it. They have to experience it.

The 8% who use AI as a sparring partner didn't learn that from a course. They saw someone else do it. Or they stumbled into it through experimentation. Or—and this is the key—they had the psychological safety to try something weird.

Here's my confession: I've been leading AI upskilling initiatives for Fortune 500s for two and a half years. And I've had to reinvent nearly every piece of the curriculum along the way.

The first thing I eliminated? Presentations. The second thing? Lead by demo, not by explanation. The third? Carve out space in every session for people to actually try the tactics themselves—not later, not as homework, but right there in the room.

These changes came from watching thousands of professionals grapple with the same question: "But how do I actually work with AI, in my work, really?" That question doesn't get answered by slides. It gets answered by doing.

The metric I obsess over now is the percent of participants who experience meaningful uplift in their actual work. Not satisfaction scores. Not completion rates. Uplift. And all of these curriculum changes are in service of moving that number up.

What Leaders Actually Need to Do

If you're responsible for AI adoption, this reframes your job entirely.

You're not in the training business. You're in the imagination business.

Model in public. Your people need to see you using AI—not just hear you talk about it. Show them what a sparring partner conversation looks like. Let them watch you iterate. Make the invisible visible. (And yes, that means being willing to look dumb in front of your team.)

The simplest tactic I know: ask "Have you tried AI?" regularly. Three things happen when you do. First, you give them the idea—often they haven't even thought to try AI for that task. Second, you grant permission—your question implies it's okay. Third, you earn the follow-up: "How did AI help?"—which stocks your reservoir of stories to celebrate.

Put a sticky note on your monitor. Ask the question even when you know the answer.

Create safety to experiment. The social penalty for AI use is real. You have to actively counteract it. Celebrate experimentation. Normalize failure. Make it safe to be seen trying.

Spark imagination before building skills. Skills follow imagination, not the other way around. Before you train on prompting techniques, expose people to what's possible. Show them the 8% behaviors. Let them see AI as a thought partner, not a search engine.

Surface the edges. Your power users are already figuring things out—but they're staying quiet. Create structures that surface and spread what's working.

One leader I know—John Devine, a global head at Meta—posts a 1-minute video to his team Slack every single week celebrating something cool someone is doing with AI. People pay attention to what leaders celebrate. That weekly drumbeat signals: this matters, I notice, keep going.

The Real Barrier

I went looking for a blocker. I found a ceiling.

The 70% who don't know "how" aren't missing information. They're missing imagination. They can't envision what a real AI collaboration looks like because they've never seen one. They're stuck in tool mode because no one showed them teammate mode. They plateau at 1-2 prompts because that's all they can imagine doing.

The path forward isn't more presentations. It's more modeling. More visibility. More sparking.

Your people are believers who are stuck. Unstick them by showing them what they've never seen.

Related: 17-Day Personal AI Bootcamp
Related: Stop Measuring AI Usage (Start Measuring AI Impact)
Related: "Have You Tried AI?" The Answer Every Leader Should Give
Related: You're Not Tired of AI. You're Scared of It.

Join over 29,147 creators & leaders who read Methods of the Masters each week.