According to Gallup, half of American workers never use AI at work

The AI Workplace Revolution Is Stalling: Gallup's 2025 Report Shows a Digital Divide That's Only Getting Deeper

For the past three years, we have been told that artificial intelligence is transforming the workplace. Consultants promised a tidal wave. Tech CEOs declared that AI literacy would soon be as essential as reading and writing. Headlines warned of mass job displacement and a future where humans compete with machines for every task. But beneath the hype, a quieter, more complicated reality has been taking shape. According to the latest Gallup "AI in the Workplace" study for the fourth quarter of 2025, the much-hyped AI revolution is not stalling exactly — it is stratifying. And the gap between those who use AI regularly and those who never touch it is becoming a chasm.

The numbers are striking. Nearly half of American workers — 49% , to be precise — report that they never use AI products in their jobs. Not occasionally. Not weekly. Never. This is not a story of Luddites smashing machines or technophobes refusing to adapt. For most of these workers, AI simply has not arrived at their desks, their factory floors, or their retail counters. And when it has arrived, they often cannot see a reason to open it.

At the same time, a smaller but significant segment of the workforce has woven AI into the fabric of their daily routines. These frequent users are developing what Gallup calls "stronger habits" — not just checking AI outputs occasionally, but relying on them for drafting, summarizing, analyzing, and even decision-making. The result is a workplace that is not being uniformly transformed, but rather split into two parallel economies: one where AI is a powerful co-pilot, and another where it might as well not exist.

The Peak Has Passed: Adoption Slows After 2024's Hype Cycle

To understand where we are, it helps to look at where we have been. AI adoption in the American workplace accelerated dramatically in late 2023 and throughout 2024. The release of increasingly capable large language models, combined with aggressive marketing from tech giants, led many to believe that a tipping point had been reached. Gallup's own data from that period showed a steep upward curve: more workers trying AI, more managers recommending it, and more companies piloting internal tools.

But the fourth quarter of 2025 tells a different story. Adoption has begun to slow. The rapid ascent of 2024 has flattened into a plateau. The workers who were going to adopt AI, by and large, have already adopted it. The remaining holdouts are not simply late to the party — they are choosing not to attend.

This slowdown is not catastrophic. The technology is not failing. But it does suggest that the early, easy gains have been captured. The "low-hanging fruit" of AI adoption — knowledge workers in tech-forward companies who immediately saw the value of automated summarization, draft generation, and data analysis — have mostly been picked. What remains is the harder work of convincing the unconvinced, integrating AI into non-digital workflows, and demonstrating utility to workers who have tried ChatGPT once, been underwhelmed, and never returned.

As Gallup's lead researcher on the study put it: "We are past the novelty phase. The question now is not whether AI can do impressive things. It clearly can. The question is whether it can do useful things for the average worker in the average job. And for nearly half of Americans, the answer so far is no."

Tech Leads, Retail and Manufacturing Lag Far Behind

The adoption divide is not random. It follows the contours of the economy itself. Technology workers — software developers, data scientists, IT support, product managers — lead the pack by a wide margin. Gallup found that 60% of tech workers report using AI regularly in their jobs. This is not surprising. Tech workers are the most likely to understand how AI works, to trust its outputs, and to have the skills to integrate it into complex workflows. For a software engineer, an AI coding assistant is not a luxury; it is becoming a baseline expectation.

But as you move away from tech, the numbers drop sharply. Finance, education, and professional services (law, consulting, accounting) form the second tier, with adoption rates in the 30–40% range. In these fields, AI is present but not universal. A financial analyst might use AI to summarize earnings reports. A teacher might use it to generate lesson plans. A lawyer might use it for initial document review. But in each case, the use is often sporadic, task-specific, and not yet embedded into daily routines.

Then come retail and manufacturing — and here the numbers fall off a cliff. Gallup found that adoption in these sectors lags so far behind that it barely registers in the aggregate statistics. Less than 15% of retail workers and barely 10% of manufacturing employees report using AI regularly. For the remaining 85–90%, AI is either unavailable, unusable, or unhelpful.

Why the gap? In retail, the core tasks are physical and interpersonal: stocking shelves, operating cash registers, helping customers find products, cleaning up spills. AI cannot do any of these things in a meaningful way. A chatbot might help a customer online, but the floor associate still has to hand them the bag. In manufacturing, the story is similar but even more acute. Factory work involves machinery, assembly lines, quality control, and safety protocols. These are not tasks that lend themselves to a text prompt or a generative model. Some factories use AI for predictive maintenance or supply chain optimization, but the workers on the line rarely see or interact with those systems. The AI is in the back office, not on the floor.

Gallup's report is blunt: "For a large segment of the American workforce, AI is not a tool. It is an abstraction. It happens somewhere else, to someone else, and its outputs arrive as spreadsheets or recommendations that feel disconnected from the physical reality of the job." This is not resistance. It is irrelevance.

The Remote Divide: Office Workers Embrace AI, On-Site Workers Ignore It

One of the most striking findings in the Gallup study involves the relationship between remote work and AI adoption. The numbers could not be more different.

Among workers in remote-capable roles — jobs that can be done from home, typically involving computers, communication, and knowledge work — 66% have adopted AI in some form, and 40% use it regularly. That is a majority. In these roles, AI is not a fringe experiment; it is a mainstream tool. Remote workers use AI to draft emails, summarize meetings, organize tasks, analyze data, and even generate creative content. The feedback loop is tight: they try something, see immediate results, and integrate it further.

By contrast, among workers in roles requiring physical presence — healthcare (non-administrative), retail, manufacturing, hospitality, transportation, construction — adoption plummets to 32% , and regular use falls to just 17% . Again, this is not about stubbornness or fear. It is about fit. A nurse cannot ask an AI to insert an IV. A truck driver cannot ask an AI to unload a pallet. A hotel housekeeper cannot ask an AI to change the sheets. The AI exists, but in a different plane, a digital realm that does not intersect with their daily physical labor.

This divide has profound implications. Remote-capable workers are gaining a powerful productivity multiplier that their on-site counterparts largely lack. Over time, this will almost certainly widen the earnings gap between the two groups. Already, remote jobs tend to pay more than on-site jobs of comparable educational requirements. AI adoption will likely amplify that difference. The worker who can produce 30% more output in the same amount of time, thanks to AI assistance, will be more valuable to their employer than the worker who cannot.

Gallup warns: "The benefits of AI are not being distributed equally. They are flowing disproportionately to workers who already had advantages — higher education, digital literacy, flexible work arrangements. This is not a critique of AI. It is a description of reality. And unless something changes, that reality will become more entrenched."

Why Non-Adopters Stay Away: "Lack of Clear Utility"

Gallup asked non-adopters a simple question: why don't you use AI at work? The most common answer, given by over 60% of respondents, was not fear, not lack of access, not concern about job security. It was something far more mundane: "lack of clear utility."

These workers have tried AI, or seen it demonstrated, or heard about it from colleagues. And they have concluded, often correctly, that it does not make their jobs easier, faster, or better. The AI might be impressive in a demo, but when applied to their specific tasks — dealing with an angry customer, fixing a broken machine, organizing a physical inventory — it offers nothing. It is a solution in search of a problem.

This is a humbling finding for AI enthusiasts. The technology is advancing at breakneck speed. Models are getting larger, smarter, and more capable. But capability is not the same as usefulness. A model that can write a sonnet about quantum physics is useless to a cashier who needs to process a return. A model that can generate a business plan in thirty seconds is useless to a welder who needs to align a joint.

The "lack of clear utility" problem is not permanent. As AI becomes more embodied — integrated into robots, smart glasses, augmented reality interfaces, and other physical-world tools — its relevance to on-site workers will grow. But that future is not here yet. And for now, the result is a stark divide: AI is a daily companion for the knowledge worker and an occasional curiosity for everyone else.

The Leadership Gap: Managers Use AI Much More Than Their Teams

Another revealing finding from the Gallup study involves hierarchy. Leaders — executives, directors, and senior managers — are adopting AI at much higher rates than the people they manage. Gallup found that 69% of leaders use AI regularly, compared to just 40% of individual contributors (non-managerial employees). That is a gap of nearly thirty percentage points.

Why are leaders so much more enthusiastic? Partly because their jobs involve tasks that AI handles well: drafting strategy documents, analyzing reports, summarizing information, preparing presentations, and managing schedules. These are all text-heavy, pattern-rich, and low-stakes enough that AI errors can be caught before they cause harm. For a manager, AI is a personal assistant that never sleeps.

But there is another factor: leaders have more autonomy to experiment. An individual contributor might need permission to install a new tool or integrate AI into a workflow. A leader can simply decide to use it. This top-down adoption pattern is not necessarily bad — leaders can model good practices and advocate for broader rollouts — but it does create a strange dynamic. The people most responsible for evaluating and promoting workers are using a tool that their workers themselves do not have or do not use. This can lead to mismatched expectations, where leaders assume AI is as available and useful to everyone as it is to them.

Gallup notes: "Leaders risk overestimating the penetration of AI in their own organizations. When they look around and see their peers using AI, they may assume that everyone is doing the same. The data suggest otherwise. In most companies, AI use is concentrated at the top and among specific functions. The average worker is far less engaged."

The Future: Disparity Will Widen, Then Competition Will Intensify

So where do we go from here? Gallup's report offers a two-part forecast.

First, the disparity will widen. The workers and industries that have already adopted AI will continue to build habits, integrate tools more deeply, and reap productivity gains. Their outputs will improve. Their value to employers will increase. The workers and industries that have not adopted AI will largely stand still — not because they are lazy or resistant, but because the technology does not yet fit their reality. This divergence will create a two-speed labor market, with AI-augmented workers pulling ahead and everyone else struggling to keep pace.

Second, and perhaps counterintuitively, the lagging industries will eventually see fierce competition. Gallup argues that when AI does become viable in retail, manufacturing, healthcare, and other on-site sectors, the adoption will not be gradual. It will be explosive. The first company that figures out how to deploy AI effectively on a factory floor, or in a retail chain, or across a hospital system, will gain a massive competitive advantage. That advantage will force everyone else to catch up quickly. The laggards will become accelerators overnight.

In other words, the calm we are seeing now — the plateau, the slowdown, the "lack of clear utility" — is not permanent. It is the lull before a different kind of storm. When AI finally crosses the chasm from digital to physical, from knowledge work to every kind of work, the disruption will be faster and more intense than anything we have seen so far. The companies and workers who prepare now, even in industries where AI seems irrelevant, will be the ones who thrive.

Gallup concludes with a warning and an opportunity: "Do not mistake the current plateau for a permanent ceiling. AI is not going away. It is going to keep improving. And one day, perhaps sooner than you think, it will arrive in your industry, your workplace, and your specific job. When that day comes, the only question will be whether you are ready."