The Doing Gap Will Kill More Companies Than AI

I saw a flashy CNBC headline over the holidays—"AI will impact jobs in 2026, say 89% of HR leaders"—and my first thought wasn't about the 89%. It was about the 11%.

Who are these leaders who don't think AI will impact jobs next year? What are they seeing—or not seeing—that the rest aren't? Are they running organizations in purely physical domains? Or have they simply made peace with irrelevance?

(I know, I know. I'm a lot of fun at parties.)

As I read the survey, I got gobsmacked by this gap: 89% of leaders believe AI will impact jobs. Yet only 67% say it's currently impacting work at their organizations. That's a 22-point gap between belief and action—between knowing something matters and actually doing something about it.

This isn't a technology problem. It's a leadership problem.

A Show of Hands Tells the Truth

I gave a keynote recently where the host warned me beforehand: "I've been hearing a lot about AI fatigue. People are kind of tired of it."

So I opened with two questions.

First: "Raise your hand if you're feeling AI fatigue." Every hand went up.

Second: "Raise your hand if you're prepared to share radical use cases you've discovered with AI." Every hand went down.

I let the silence sit for a moment. "That's kind of ironic, right? We're all tired of talking about it, yet none of us are doing it."

The CNBC survey shows the same pattern at scale. Leaders believe AI matters—89% of them say so. But believing and doing are very different things. One HR leader in the survey admitted: "It is still too early to tell. Most organizations have not completely integrated AI tools into every job, nor have they identified ways to measure the impact."

Too early to tell? We're three years into the generative AI era. The London School of Economics found that employees using AI save 7.5 hours per week—nearly a full workday. That's not "too early to tell." That's choosing not to look.

The Line That Should Terrify You

Real talk? This is the line from the survey that astounded me:

When asked why they're planning workforce reductions, HR leaders cited only "general need to cut costs"—not inflation, not tariffs, not decreased demand, and not efficiency gains from AI.

Read that again. (I'm serious.)

Organizations are cutting people while simultaneously admitting that AI isn't making them more efficient. These are the same leaders who believe AI will impact jobs next year. They just haven't captured any of that impact yet.

If you're cutting headcount but can't point to AI-driven efficiency gains, you're not leveraging AI—you're just shrinking. And shrinking without transformation isn't a strategy. It's managed decline.

A Claim I Don't Believe

The survey found that 78% of leaders say AI has made their workforce "more innovative."

I'd love to believe that. But my research suggests the opposite.

I served as co-researcher on a study examining whether teams using AI to generate creative solutions came up with better ideas than unassisted teams. What we found was troubling: the vast majority of teams with AI access actually underperformed those without access on standard creativity measures, as judged by third-party objective evaluators.

Fewer bad ideas, sure. But also fewer great ideas. AI produced lots of average.

(Which, when you think about it, makes sense. Large language models are statistical models. They're literally designed to predict the most probable next word. "Most probable" and "most creative" aren't the same thing.)

The few teams that outperformed? They were exceptional in one specific way: they treated AI like a teammate, not a tool. Multi-turn conversations. Giving feedback. Pushing back. Pulling for more. It was a completely different mode of interaction than the shallow, superficial prompting that produced mediocre outputs for everyone else.

So when 78% of leaders say AI made them more innovative, I suspect what they mean is: "We're producing more stuff faster." That's efficiency, not innovation. And if you're producing mediocre stuff faster, you're not winning—you're just losing more efficiently.

The 7.5-Hour Question Nobody's Asking

The London School of Economics study found that employees using AI save an average of 7.5 hours per week—nearly a full workday.

That's remarkable. But who’s asking: What are people doing with those 7.5 hours?

Are they pursuing ideas they never had capacity for? Or are they doing the same work with fewer people?

Jensen Huang framed this perfectly: "The real question is, are you a hopeful person, an optimistic person who believes in idea creation, or are you somebody who believes there are no new ideas left and quite frankly we're just working?"

Organizations that use those 7.5 hours to pursue their backlog of impossible ideas will transform industries. Organizations that use those hours to cut headcount will be transformed by their competitors.

What CNBC’s Survey Actually Means

If 89% of HR leaders believe AI will impact jobs next year, that's not a warning. It's an invitation.

An invitation to stop making peace with pain. (The CEO who told me he had "no pain points to automate" also mentioned he spends six hours editing every report his team sends to clients. When you normalize inefficiency so completely, you stop seeing it as an opportunity.)

An invitation to punish inaction, not just failure. Good companies reward success and punish failure. Great companies reward success and failure, and punish inaction. If someone on your team runs ten AI experiments and all ten fail, that's better than zero experiments.

An invitation to model the behavior you want to see. Your team won't use AI until they see you using it. When someone asks you a question in a meeting, try this: pull up ChatGPT on the shared screen and say, "Before providing an answer, please ask me any questions you need to better understand this situation." Then work through the answer together. Make your AI thinking visible.

An invitation to measure better use, not more use. Stop tracking "number of AI logins" and start asking: "Which workflows have been fundamentally transformed?" More use isn't better use. Better use is better use.

An invitation to commission yourself. Don't wait for your organization to figure this out. Jeff Bezos saw Napster and told Steve Kessel: "Your job is to kill your own business." Kessel invented the Kindle. What would it look like to commission yourself to transform your own role before someone else does?

The 11% Who Don't See It Coming

Let's come back to that 11%.

These aren't jobs that are immune to AI. These are leaders who don't believe AI will impact jobs at all. In 2026. After three years of the generative AI era.

I suspect they fall into three categories: leaders running organizations with genuinely zero knowledge work, leaders who will be replaced by leaders who do see it coming, and leaders who've opted out of paying attention—and will discover too late that the world didn't wait for them.

89% of HR leaders believe AI will impact jobs next year. They fall in two buckets. The question is, which bucket are you in? Are you among those leaders are closing the gap between belief and action—or just nodding along while the doing gap gets wider.

Try This Now:
Look at your calendar from the last three days. For each meeting, task, and decision, ask yourself: "Did I use AI?" Not "Could I have?" but "Did I?"

If you're honest, and you're below 50%, you're in the gap. You believe AI matters. You're just not doing anything about it.

And every day you stay there, the gap gets wider.

Related: It's Not AI Fatigue. It's AI Insecurity.
Related: Stop Being a Hypocrite

Join over 29,147 creators & leaders who read Beyond the Prompt each week.