Artificial intelligence has never been more capable. Oddly enough, people have never seemed more uneasy about it.

That wasn't supposed to happen.

For the better part of two years, AI has been marketed as the ultimate productivity breakthrough. It can summarize meetings, draft presentations, analyze data, write code, answer questions, and perform in seconds tasks that once consumed entire afternoons. By nearly every measurable standard, the technology has improved dramatically.

Yet public sentiment is moving in the opposite direction.

According to recent research from Jobs for the Future (JFF), Americans became significantly more pessimistic about AI between 2024 and 2025, with growing concerns about its impact on jobs, wealth creation, and overall quality of life. At the same time, confidence that employers are adequately preparing workers for an AI-driven future declined sharply.¹

Do you believe AI is having more harm than good, or more good than harm on people’s ability to find jobs, build wealth, and secure their quality of life, in society as a whole?

That reversal is fascinating because it breaks the normal pattern of technological adoption.

Most innovations become less intimidating as people grow familiar with them. The internet looked scary until it became indispensable. Smartphones seemed unnecessary until we started reaching for them before our morning coffee. Even online banking eventually became so routine that few people think twice before depositing a check from their kitchen table.

AI is proving different.

The more exposure people have to it, the more conflicted many seem to become.

Part of that anxiety is easy to understand. Every transformative technology creates winners, losers, and a healthy amount of uncertainty. Whenever a machine learns to perform a task once reserved for humans, people naturally begin wondering where they fit into the picture.

But I suspect something deeper is happening.

The emerging concern isn't merely economic.

It's cognitive.

People aren't just worried AI will replace portions of their work. Increasingly, they're wondering what happens when it replaces portions of their thinking.

That concern moved from theory to research when a study from MIT Media Lab examined how people performed writing tasks with and without AI assistance. Researchers found that participants relying heavily on ChatGPT demonstrated lower brain connectivity, weaker recall, and less ownership over the ideas they produced. The study introduced a memorable phrase: cognitive debt.²

It's one of those concepts that immediately clicks.

We understand financial debt because we've all borrowed against tomorrow.

Cognitive debt suggests we may be doing something similar with our thinking.

The appeal is obvious. Why spend an hour organizing ideas when AI can produce a polished outline in ten seconds? Why wrestle with a difficult first draft when a chatbot can generate one instantly? Why spend thirty minutes researching a topic when a summary appears with a single prompt?

The convenience is extraordinary.

That's precisely why the concern is worth discussing.

Human beings develop expertise through effort. We build judgment by struggling with ambiguity. We learn by making mistakes, revising conclusions, and occasionally staring at a blinking cursor while our brains do the difficult work of connecting ideas.

AI short-circuits some of that process.

Not because it's flawed.

Because that's exactly what it was designed to do.

The irony is that the same technology increasing productivity may also be reducing the mental friction that helps develop competence in the first place.

That tension helps explain why so many workers report seemingly contradictory feelings about AI. They appreciate what it does while simultaneously feeling unsettled by what it might be doing to them.

A separate study published in Societies (MDPI) found a relationship between frequent AI use, cognitive offloading, and weaker critical-thinking performance.³ Again, the researchers were not arguing that AI makes people less intelligent. Their concern was subtler. When people consistently outsource mental tasks, they may exercise those cognitive muscles less frequently.

Anyone who has followed directions exclusively through GPS understands the phenomenon.

Most of us arrive at the destination successfully.

Far fewer could explain how we got there.

Technology has always altered human behavior in ways both obvious and invisible. The calculator changed arithmetic. Search engines changed memory. Social media changed attention spans. AI may prove to be the most consequential cognitive technology yet because it doesn't merely store information or deliver information.

It helps generate thought itself.

That distinction matters.

The contradiction becomes even more apparent when you look at workplace behavior. AI adoption continues to rise across industries, with Gallup reporting increasing usage among knowledge workers and managers.⁴ At the same time, a Reuters/Ipsos poll found that a majority of Americans remain concerned about AI's impact on jobs and society.⁵

Put differently, people are embracing the technology while simultaneously worrying about where it leads.

That tension helps explain why sentiment has deteriorated even as usage accelerates.

Meanwhile, employers are increasingly rewarding AI proficiency. Microsoft's Work Trend Index found that organizations are placing growing value on AI skills, while PwC's AI Jobs Barometer suggests workers with AI capabilities are seeing stronger demand and faster wage growth.⁶⁻⁷

The market is effectively sending two messages at once.

Learn AI.

And be careful what happens if you rely on it too much.

The first phase of the AI boom focused almost entirely on what the technology could accomplish. Investors measured productivity gains. Executives chased efficiency. Software companies competed to release increasingly impressive capabilities.

The second phase may focus on a more uncomfortable question:

What happens to human cognition when intellectual effort becomes optional?

That question doesn't have an obvious answer, and it certainly doesn't justify the more apocalyptic predictions surrounding artificial intelligence. History is littered with examples of society overestimating the short-term dangers of transformative technologies while underestimating their long-term impact.

But it does help explain why sentiment is deteriorating even as adoption accelerates.

People aren't losing confidence in the technology.

They're wrestling with what the technology means for them.

After all, confidence rarely comes from having answers handed to us. It comes from earning them. The lawyer develops confidence by building arguments. The advisor develops confidence by solving complex client problems. The marketer develops confidence by staring at a blank page and eventually finding something worth saying.

There is a subtle but important difference between using AI to enhance thinking and using AI to replace it.

One creates leverage.

The other creates dependence.

The winners of the AI era will almost certainly be those who learn to recognize the difference.

Because the greatest risk may not be that artificial intelligence becomes smarter than humans.

It may be that humans become less willing to exercise the intelligence they already possess.

Related: Affluent Clients Are Craving Something Most Brands No Longer Deliver

Sources

¹ Jobs for the Future. Worker Anxiety Over AI Is Growing and Employers Aren't Preparing Employees for What's Next (2025).

² MIT Media Lab. Your Brain on ChatGPT: Accumulation of Cognitive Debt When Using an AI Assistant for Essay Writing Tasks (2025).

³ Gerlich, Michael. AI Tools in Society: Impacts on Cognitive Offloading and Critical Thinking. Societies (2025).

⁴ Gallup. Rising AI Adoption Spurs Workforce Changes (2025).

⁵ Reuters/Ipsos Poll. Americans' Concerns About AI and Employment (2026).

⁶ Microsoft. Work Trend Index (2026).

⁷ PwC. AI Jobs Barometer (2025).