If your job feels like it is changing faster than anyone will explain, you are not alone, and the numbers suggest your employer may be losing more than goodwill. Gallup’s latest read on the US workforce shows engagement flat at a low level while companies push artificial intelligence into daily work.
In this article we will discuss what the figures say, how clear AI expectations relate to engagement, what the privacy backlash looks like, and what employees and managers can do about it.
Engagement sits at 31 percent and refuses to move
Gallup’s survey covering the first half of 2026, as reported by dpa and The Star on 12 August, found that 31 percent of US workers are engaged in their jobs. That is unchanged from last year and five percentage points below the level in 2020. Gallup estimates the shortfall amounts to roughly 8 million fewer engaged American employees than six years ago.
Those figures are worth sitting with. A flat number can look like stability, but it means the workforce has not recovered from the slide that followed the pandemic years, and that most people are doing their jobs without much enthusiasm for them. Engagement is not the same as satisfaction or loyalty. A person can be fairly content with a paycheck and a schedule and still be disengaged in Gallup’s sense.
The data comes from the United States, and the picture in other countries differs. Gallup has reported on engagement worldwide for years, with wide variation by region, so readers elsewhere should avoid assuming a 31 percent figure applies to them. What travels better is the underlying logic: how work is organized and communicated has a strong effect on how people feel about it.
Why a flat number still matters
Flat engagement during a period of rapid technological change is a warning sign. Employers often introduce new tools while their workforce is already stretched, and the odds of a smooth transition are lower when people are not invested in the outcome. If the baseline is weak, a poorly handled AI rollout can push it lower.
Who is affected
Nearly everyone with a job in the United States sits somewhere in this picture, but the experience differs by role. Frontline staff who are told to use a new tool with little notice feel the strain most directly. Middle managers are often caught between targets set above them and doubts expressed below. Small businesses face a different problem, since they rarely have a dedicated team to write policy and may be copying whatever a software vendor suggests.
Defining engagement beyond the buzzword
Gallup defines engagement as “employees’ involvement in and enthusiasm for their work and workplace.” Both halves of the phrase matter. Involvement means people feel connected to what they are doing and know how it fits the larger goal. Enthusiasm means they care enough to put in effort that goes beyond a minimum.
It helps to separate engagement from a few concepts it is often confused with. Happiness is a mood, which can rise and fall in a week. Satisfaction is a judgment about conditions such as pay and benefits. Engagement is closer to commitment and energy, and it tends to depend on everyday experiences: whether the manager sets clear goals, whether the work uses a person’s strengths, and whether someone notices good work.
The drivers that rarely change
Decades of workplace research, from Gallup and many others, point to a familiar list of drivers. People want to know what is expected of them, to have the tools to do the job, to receive recognition, and to feel that their opinions count. None of these is a mystery, and none requires new technology. Notice that the first item, clarity about expectations, is exactly the one the AI findings turn on.
Why it affects the business
Engagement matters to employers because engaged teams are generally associated with better retention and productivity. The exact size of the effect varies by study and industry, so treat any specific promise skeptically. The safer statement is that disengagement is costly in ways that rarely show up in a single report: more turnover, more mistakes, and slower adoption of anything new.
The long road since 2020
The five-point gap against 2020 is a reminder that engagement is slow to build and quick to lose. Between then and now, workplaces went through remote and hybrid arrangements, waves of layoffs and hiring, and changing expectations about flexibility. Each of those shifts asked people to adapt, and trust took hits along the way. AI arrives on top of that history, which is why the way it is introduced carries so much weight.
Clear AI rules go with higher engagement
The most useful finding in the report concerns how companies introduce AI. According to the coverage, Gallup found that employees are “substantially more likely to be engaged” when their company sets “clear expectations” and a “thoughtful implementation plan” for AI. In other words, it is not just whether a business adopts the technology that matters, but how it explains and manages it.
That fits what many workers say informally. Staff are rarely opposed to a tool simply because it is new. What worries them is ambiguity: whether using it is required or optional, whether the output will be judged differently, whether their role is shrinking, and who is responsible when a machine-generated answer is wrong. Left without answers, people guess, and the guesses tend to be pessimistic.
A caution on interpretation applies. The reporting describes an association, not proof of cause and effect. Companies that plan carefully may also be better managed in other ways, and those other strengths could account for some of the higher engagement. Still, the pattern is plausible, and the remedy it suggests costs little: tell people what is expected and why.
It helps to keep the wording human. A policy full of legal phrasing invites people to ignore it, while a plain page that says what is encouraged, what is forbidden and who to ask tends to get read. Review it every few months, because tools and risks change quickly and last quarter’s rules may no longer fit.
What clear expectations look like
In practice, clear expectations answer a small set of questions in plain words. Which tasks are staff encouraged to use AI for? Which are off limits, such as entering customer data or confidential documents into outside tools? How should AI-assisted work be checked and labeled? Who can people ask when a case is unclear? A short written policy that addresses these points is more helpful than a long document no one reads.
What a thoughtful implementation plan looks like
A plan goes beyond a policy. It describes the order in which tools will arrive, which teams will try them first, how training will work, and how feedback will shape the next step. A thoughtful plan treats the first group of users as a pilot and expects to learn from it. It also admits what is not yet decided, which is more credible than pretending everything is settled.
The role of middle managers
Rollouts rise or fall on the people who translate company strategy into daily routines. A senior leader can announce an AI initiative in a single email, but the manager is the one who explains what it means for Tuesday’s workload. When managers are not briefed or trained themselves, they cannot answer basic questions, and their teams read the silence as a sign that no one is in charge. Investing in manager preparation before staff rollout is one of the cheapest ways to avoid that outcome.
Why speed is not the same as progress
There is also a lesson here about pace. Some firms rush to announce that they are an AI-first business because they fear being left behind. The risk is that the announcement outruns the preparation. Staff then absorb pressure to use the tools without the guidance to do so well, and the early results disappoint.
A slower start with clear rules often reaches useful results sooner, because fewer people need to be won back after a bad first impression.
The two trillion dollar productivity claim needs caution
The coverage also reported that Gallup links productivity losses to the AI push, with an estimated cost of around $2 trillion a year. That is a very large number, and it should be handled carefully. The reporting presents it as Gallup’s assertion, and the method behind it is not explained in the article, so readers cannot judge how it was calculated or what is counted.
There are reasonable ways to read it. One is that disengaged or confused employees do less valuable work, and that when millions are caught in a badly managed transition, the lost output adds up across the whole economy. Another is that the number is an extrapolation resting on assumptions that may or may not hold. Without the method, neither reading can be confirmed.
How to treat a headline number
A sensible habit with large figures is to ask three questions. What is being measured? Over what period and for which population? And what is the comparison point, meaning lost relative to what? If a report does not answer these, treat the figure as a signal of direction rather than a fact to quote. That is the right stance here: the direction, that confusion about AI can cost productivity, is believable, while the exact amount is unverified.
Training is part of clarity
Many workers say they have been handed AI tools without lessons on using them. Training does not have to be elaborate. A one-hour session with real examples from the team’s own work, followed by a place to ask questions afterward, goes a long way. Pair it with a plain statement of what quality means, so that people know they will not be penalized for catching and fixing a machine error. In a good rollout, careful checking is treated as part of the job, not as a sign of resistance.
Privacy worries and the missing opt-out
A separate survey, led by King’s College London in July, adds another piece. It found that 42 percent of participants said they had started limiting their use of chatbots because of privacy concerns. Participants also complained about a lack of opt-outs at work, meaning they felt they had little choice about whether AI tools were used in their jobs.
The sample and country coverage of that survey are not detailed in the reporting, so it is best read as an indication of sentiment rather than a measure of the whole workforce. Even so, it points to a tension that many employers underestimate. Workers may be asked to use tools that record prompts, store documents or monitor performance, while having limited information on where that data goes.
Why privacy shapes engagement
Trust is a core ingredient of engagement. If staff suspect that a tool is quietly assessing them, or that sensitive information typed into a chatbot might be seen or reused, they may avoid it or use it half heartedly. That damages the productivity the tool was meant to deliver. Clear rules about data, retention and monitoring are therefore not a legal nicety but a practical part of adoption.
Rules differ by country
Data protection law varies widely. Some places, including the European Union and the United Kingdom, have broad rules that govern how employers handle personal data and may require notice or consultation. In the United States, protections for workers are patchier and depend on the state and the situation. Readers should check their own jurisdiction and the official guidance, because this article offers general information only and cannot judge a specific workplace.
What it means for trust at work
Taken together, the evidence suggests that the conversation about AI at work is not only about technology. It is about whether people believe their employer is being straightforward. Workers who feel informed tend to treat a new tool as an aid.
Workers who feel left in the dark tend to treat it as a threat, and no amount of enthusiasm from the executive floor changes that. That is why the practical fixes are mostly about communication, not software.
Employees can ask for specifics
Workers have more influence than they sometimes assume, especially when questions are phrased constructively. The aim is to get clarity, not to block a tool.
Ask for the policy in writing
If your employer encourages AI use, ask whether there is a written policy and where to find it. If none exists, asking is useful in itself, since it signals that staff want guidance. A manager who cannot answer may pass the question upward, which is how many policies begin.
Ask the privacy questions
Good questions are specific. Which tools are approved? What data may be entered? Are prompts stored, and by whom? Is any AI output used to evaluate staff? You are entitled to ask these, and a clear answer builds trust. If you are not satisfied, avoid entering personal, customer or confidential information until you have an answer.
Ask about skills and training
Use the moment to ask how you will be trained. Learning to work with AI tools is a skill that can make you more valuable, and it is reasonable to request time during work hours to build it. Keep notes on what you learn, since that also helps in later reviews and job searches.
Peer support also counts. Colleagues often share the same worries, and raising an issue as a team can be easier than doing it alone. A staff group or informal meeting can collect questions and present them together, which gives managers a clearer picture of what needs addressing.
Raise concerns early and calmly
If something troubles you, such as an unrealistic expectation about output or a worry about monitoring, raise it with a manager, human resources or a staff representative, depending on your workplace. Describe the specific problem and a suggested fix. Record dates and conversations in case you need them later. For serious legal concerns, speak with a qualified professional, since this article cannot advise on individual cases.
Be honest about limits and mistakes
AI tools make errors, sometimes confident-sounding ones, and employees need to know that reporting a problem is welcome. If the first person to flag a flawed output is treated as a nuisance, the next person will stay quiet, and flaws will spread into reports and customer messages. Encourage staff to share examples of where the tool failed as well as where it helped. That builds a realistic picture of what the technology can do and keeps standards high without fostering fear.
Managers set the conditions for trust
For managers, the Gallup finding is practical. The difference between a confusing rollout and a clear one is largely within a manager’s control, and it begins with talking to the team.
The table below sets out a basic checklist that follows the findings and ordinary good practice. It is a suggested starting point rather than a formula from the survey.
| Step | What to do | Why it helps |
| Set expectations | State which tasks AI is for, which are off limits and how output is checked | Reduces guessing and fear about hidden rules |
| Explain the plan | Share the order of rollout, timing and who is involved | Shows the change is considered, not rushed |
| Pilot first | Start with a small volunteer group and collect feedback | Finds problems before they reach everyone |
| Train | Provide time and resources to learn the tools | Builds confidence and skill |
| Protect privacy | Say what data is stored, who can see it and what is monitored | Supports trust and willing use |
| Listen | Create a regular channel for questions and concerns | Keeps the plan responsive |
Do not hide the hard parts
Many leaders avoid discussing job security because they have no certain answer. Silence is usually worse. It is better to say what is known, what is not, and when updates will come. People tolerate uncertainty better than evasion.
Recognition matters too. When someone finds a smart way to use a tool, thank them publicly and ask them to teach the team. Engagement grows when people feel their expertise is valued, and AI adoption tends to spread faster through trusted colleagues than through directives.
Offer choices where you can
The King’s College finding about missing opt-outs suggests that giving staff some control can ease resistance. Where tools are optional, say so. Where they are required, explain why and describe what safeguards exist. Even a small degree of choice, such as which tasks to try first, can change how a rollout feels.
Measure what matters
A useful measure is simple and regular. A monthly pulse of two or three questions, such as whether staff feel clear on the rules and whether the tools save them time, gives an early warning long before an annual survey would.
Track whether the tools are helping, not only whether they are being used. Ask teams if they save time, where quality drops and what causes frustration. Treat the answers as information, not criticism, and adjust accordingly. A rollout judged only by login counts will tend to produce people who log in without any benefit.
A short illustration
Consider an illustration with invented round numbers. A company of 200 people announces a new AI writing assistant with a one-line email. Within a month, perhaps a quarter of staff use it freely, another quarter avoid it because they fear the output will be criticized, and the rest use it secretly and do not tell anyone.
Now imagine the same company holds a short briefing, publishes a one-page rule sheet, names ten volunteers to try it first and asks them to report back in a month. The tool is identical, but the second approach gives everyone the same information, and that shared footing is what the Gallup finding points to.
Where to start on Monday
Whether you are an employee or a manager, the first moves are modest. They are also unlikely to be wasted, because they rest on drivers of engagement that have held for years.
Employees can begin by finding out what the rules are. Read what exists, ask what does not, and note any gaps in privacy or training. Then try one low-risk use of an approved tool and judge for yourself whether it helps. Share what works with colleagues, since peer examples are often more convincing than a company memo.
Managers can start with a one-page statement of expectations and a short listening session with the team. Name the pilot group, pick a review date and tell people what will happen next. Resist the urge to announce everything at once, because a staged plan is easier to adjust.
Finally, keep perspective on the evidence. The engagement figure is a single survey, the productivity estimate is unexplained, and the privacy survey is a snapshot. Together they support a modest, defensible conclusion: when people know what is expected and trust how the tools are used, they respond better. That is a low-cost fix worth trying before anything more elaborate.
Conclusion
The coverage is a news summary of survey findings, so some detail is missing: the exact wording of the questions, the number of respondents, and how the productivity estimate was derived. Readers who need to rely on the research for a decision, such as a company designing a policy, should look at the published Gallup materials directly and check for newer releases, since figures may have been updated.
In the meantime, the advice here rests on long-standing engagement research and common sense rather than on any one number.
Frequently Asked Questions
What percentage of US workers are engaged at work?
According to Gallup’s survey of the first half of 2026, 31 percent of US workers are engaged. That is unchanged from last year and five points below 2020.
How does Gallup define employee engagement?
Gallup describes it as employees’ involvement in and enthusiasm for their work and workplace. It differs from satisfaction, which is more about conditions like pay.
Do clear AI rules really improve engagement?
Gallup found employees are substantially more likely to be engaged when companies set clear expectations and a thoughtful implementation plan for AI. The finding shows an association, not proof of cause.
How much does the AI push cost in lost productivity?
Gallup is reported to estimate around $2 trillion a year, but the method was not explained in the coverage. Treat it as an unverified estimate.
Are workers worried about AI privacy?
A King’s College London survey in July found 42 percent of participants had started limiting chatbot use over privacy concerns. Some also complained of no opt-outs at work.
What should I ask my employer about AI at work?
Ask for the written policy, which tools are approved, what data may be entered, and whether prompts or output are monitored. Also ask about training and who to contact with concerns.