It seems like there’s a story being told in just about every workplace right now. It goes something like this: the future belongs to the fastest AI adopters. The people who pick up the new tools first, build the slickest workflows, automate the most of their job, are the ones who’ll thrive in the next decade. Everyone else will be left behind.

Shopify’s Tobi Lutke, perhaps most famously, has called embracing AI a “fundamental expectation” ( as reported by CNBC). 

It’s a clean narrative. We’d say it’s also turning out to be incomplete.

Perhaps, the people who are actually doing well in this new environment don’t look much like the breathless adoption story would predict. Perhaps, they’re not necessarily the ones with the most prompts saved or the cleverest automation chains. A lot of them aren’t even early adopters in any meaningful sense.

What they have in common is something quieter. They think clearly. They engage with hard problems. They notice when an AI output is plausible but wrong. They make decisions that age well, not just decisions that ship fast.

In other words: critical thinking is still outperforming fast adoption.

And we’d argue data is starting to back that up in ways most people haven’t fully absorbed yet.

What the market is actually paying for

The WEF Future of Jobs Report 2025 ranks analytical thinking as the most sought-after core skill for 2025, with 7 in 10 employers considering it essential. It’s been the top of the list for two editions (2023 and 2025) in a row.

The hype cycle around AI hasn’t shifted that. If anything, it’s reinforced it.

AI and big data have surged up the rankings, sure. They’ve climbed from 15th place in the 2023 report into the 11th in the core skills list in 2025. That part of the adoption story is real.

But here’s what gets less attention. Over the same two-year window, leadership and social influence climbed  in importance. Resilience, flexibility and agility climbed. Creative thinking is one of the top skills. Curiosity and lifelong learning sits in the top 10.

What’s actually happening is both things at once. AI fluency is rising and the human thinking skills are rising right alongside it. The market is asking for both, and the professionals who only have one half of that equation are the ones quietly losing ground.

Fast adopters who skipped the thinking part are probably producing more output than ever. But is that output translating into better outcomes? 

What gets lost when adoption races ahead of thinking

The Harvard Business School study with Boston Consulting Group, Navigating the Jagged Technological Frontier, captured this dynamic with unusual clarity. The researchers gave consultants AI access on realistic work tasks. Inside the zone where AI was capable, “AI significantly improved performance and quality for every model specification, increasing speed by more than
25%, performance by more than 30%, and task completion by more than 12%”. 

But here’s the part most people don’t quote. On tasks that fell outside what AI could actually do well, consultants using AI were 19 percentage points less likely to produce correct solutions. The tool didn’t just fail to help. It may have actively misled them, because they trusted it on problems where they shouldn’t have.

Plenty of fast adopters in that study probably produced confidently wrong work. Plenty of slower, more careful users produced better outcomes by knowing when to lean on the tool and when to lean on themselves.

That’s the dynamic playing out across knowledge work right now, just at a much larger scale.

The cognitive cost the productivity gains hide

There’s another piece of this that doesn’t show up in the productivity numbers, and it’s worth taking seriously.

A 2025 Microsoft Research study of knowledge workers found that workers with higher confidence in generative AI exerted less critical thinking effort, while those with higher self-confidence in their own abilities exerted more.

It seems, the more people trusted the tool, the less they engaged with the work themselves. The cognitive effort shifted from doing the task to verifying the output. 

How to get more of this in your own work

A few practical moves if you want to invest in critical thinking rather than racing on adoption.

First, run a weekly audit. Where did AI help? Where did it hurt? Where did you reach for it before you actually needed it? Most people never look back at how they used these tools, which means they never learn the pattern of where their own judgment beats the model’s.

Second, build a “no tools” block into your week. Two or three hours where you work on something hard without opening any AI. The goal isn’t to be a Luddite. It’s to keep the muscle of independent thinking from atrophying. Use what you produce in that block as the foundation, then bring tools in afterwards.

Third, get specific about what you actually own. The skills employers are paying premiums for (judgment, originality, ability to read complex situations, ability to make hard calls) get built one situation at a time. 

The bottom line

It would be easy to say that fast adoption is the path to thriving. 

But we’d say the truth is messier and more durable. The people who’ll thrive in an AI-driven world are the ones who can think clearly when problems are hard, recognize when a tool is leading them astray, and bring real judgment to work the model can’t fully understand.

Adoption is part of the picture, but it isn’t the whole picture.