AI Saves Time. 66% Of Employees Stay Online To Hide It

Initially published on Forbes August 2, 2026

AI is freeing up employees’ time. Companies are giving them every reason to hide it.

Leaders say employees must transform how they work. Yet they continue to reward those who appear fully occupied by current work. Those expectations are incompatible.

AI transformation requires capacity. A company with no available capacity cannot adapt. It can only continue executing what it already knows how to do. We understand this in other parts of the business. Organizations maintain financial reserves because unexpected events occur. Technology teams build spare capacity into critical systems because operating continuously at the limit creates fragility.

When it comes to people, many companies still assume that every available minute should be filled. That approach leaves no room for learning, experimentation, reflection or transition. It also encourages employees to conceal the time they have saved, because making that capacity visible will only cause it to disappear.

The result is a strange organizational performance: leaders invest in AI to create efficiency, employees use AI to complete work faster, and then everyone collaborates in maintaining the appearance that the work still requires the same amount of time.

What Happens When Efficiency Creates More Work?

AI reduces the time employees need to complete existing work. If the organization responds by filling every newly available minute with another task, employees quickly learn that there is no benefit to becoming more efficient. Saving an hour does not create an hour they can use differently. It simply produces another hour of assigned work.

In the old employment model, employers effectively bought employees’ time. That is why many organizations assume that when technology completes part of the work, the time it releases must immediately be filled with more work.

The rational employee response is to hide the hours saved and continue to “look busy.” Which is exactly what employees seem to be doing.

In a new Software Finder survey, 66% of respondents in the U.S. said they stay online or appear active after completing their work. Workers who did this reported spending an average of almost five hours a week maintaining the appearance of productivity. The finding that should worry managers even more: 64% of respondents said they had deliberately slowed down their work to avoid finishing too early because completing tasks quickly created additional expectations.

In other words, employees these days are hiding their efficiency. Instead of using the time to learn a new skill, rethink a process or prepare for what the business will need next, they move their mouse. They leave a document open, delay responding to a message so they will still appear busy later. The Software Finder survey also found that among employees working for companies that use productivity-monitoring tools, 63% said monitoring made them more likely to fake activity.

The pattern runs up the org chart too. The same survey found that 73% of managers admit to faking productivity for their own bosses. Whatever produces this behavior in employees is producing it in the people who manage them as well.

And when respondents were asked what they would do if finishing early carried no consequences, 71% said they would log off immediately. A workforce given permission to stop when the work is done will mostly stop, rather than manufacture more hours to look occupied. People are responding rationally to what the organization rewards, at every level of the chart.

The U.S. findings are not unique. An Indeed survey of hybrid office workers in Germany found that two-thirds had taken deliberate steps to appear more productive or engaged. More than a quarter had artificially maintained an active online status, while 56% believed their employer valued presence more than measurable results.

That is what happens when a measure becomes the objective. People learn to produce the signal the organization is looking for.

If presence is measured, they remain visible. If messages are measured, they send messages. If completed tasks are measured, they divide work into more tasks. If AI consumption is measured, they consume more tokens.

The company gets more activity. It does not necessarily get more value.

Why AI Productivity Metrics Can Misread Employee Value

The same proxies that push employees toward busywork can also enter more consequential decisions about performance and employment. That risk surfaced in the recent lawsuit filed by 26 Meta employees over the company’s layoffs.

The employees allege that Meta used internal AI systems, activity data, AI-token-usage dashboards and algorithmically assisted performance information when selecting people for layoffs. They claim that employees on protected medical, parental or family leave could not accumulate the same signals and were therefore disadvantaged.

Meta denies the allegations, saying that people made the workforce decisions using documented, neutral criteria and that AI did not select employees for termination. The case has not established that the disputed systems determined the layoff list. It does raise a broader question regardless of its legal outcome: what information are managers learning to associate with valuable work?

If managers receive a list already organized around visible activity, recorded output or AI usage, the measurement system has defined what counts as contribution before the manager begins reviewing it. A person can override an individual result, but that person is still working inside a picture of employee value created by the available data.

Keystrokes show that someone typed. AI-token consumption shows that someone used AI. Neither tells a company whether AI improved the outcome, saved time or generated unnecessary work. A high volume of messages does not demonstrate influence. A full calendar does not prove contribution. A large digital footprint may indicate productivity, but it may also indicate inefficient processes, unnecessary meetings or work that should not have been done.

Someone can produce a great deal of visible output while solving the wrong problem. Someone else can prevent an expensive mistake in a short conversation that leaves almost no measurable trace.

Companies that mismeasure the work eventually cause employees to change their behavior to produce whatever the company has decided to measure.

How Companies Should Use The Time AI Saves

AI was supposed to create room for more valuable work, not more of today’s work squeezed into fewer minutes. Getting there means slowing down to speed up.

Employees need time to learn how their jobs are changing. They need to experiment with new tools, understand where those tools fail and redesign the processes into which the tools are being introduced.

They also need time to build the skills required for work the organization does not yet perform. The World Economic Forum’s Future of Jobs Report found that employers expect nearly 40% of the skills required at work to change by 2030. The skills gap is already the most frequently cited obstacle to business transformation among the employers surveyed. Out of every 100 workers, the report estimates that 59 will need training by 2030.

Where will that learning happen if organizations fill every minute AI saves with additional output?

Some of the time freed up by AI must also remain available for work that does not produce an immediate, easily counted output: improving quality, serving customers and addressing problems that were previously neglected.

Measure What The Organization Needs Next

Organizations can create just as much busywork by measuring the wrong outputs as they can by measuring the wrong activity. What matters are outcomes: whether the work moved the organization toward its goals. Those outcomes also depend on future capability, the ability of the organization and its people to do what will be required next.

It’s easy to neglect future capability because its value is exactly that: in the future. It will not appear during the next financial cycle. Learning a new system, redesigning a workflow or developing judgment may temporarily reduce visible output. But without that investment, the company may become highly efficient at work that is becoming irrelevant.

Before increasing expectations in response to AI, leaders should decide what they want to do with the capacity AI creates.

How much should improve customer outcomes? How much should increase quality? How much should be used to redesign work? How much should be invested in learning, experimentation and the skills the business will require next?

If leaders do not answer those questions, the easiest response will always be to add more tasks.

And employees will continue learning how to look busy.