You attend the training. You learn how to use the AI tool your organization has introduced. You try it on the work you already do, and some of it really does move faster.
But you are still working through lunch. Your inbox is still full. The assignments keep coming, and now there is an expectation that you can turn them around sooner.
You did what you were asked to do. You learned. You adapted. So where is the relief?
That is the question I want us to take seriously. An experienced professional who asks it deserves a better answer than another reminder to embrace change.
AI is being presented as a way to make work easier. Yet a person can become more capable with the technology and feel more stretched by the job. When the promise and the working day look that different, it can feel like smoke and mirrors.
Why does this happen?
Because saving time on a task does not determine what happens to that time. The organization’s expectations, the amount of work coming in, and the responsibility for getting it right all shape the answer.
The saved time is already spoken for
Imagine being told that a report which used to take an afternoon can now be drafted in an hour. That sounds useful. There may be other work you have been struggling to get to. Perhaps you could finally finish the day without carrying an unfinished task into the evening.
Then the requests change. Since the first version is faster, could you prepare another version for leadership? Add a separate summary for the client? Have both ready before the meeting?
This is an illustrative example, but it exposes the problem: the hour saved never becomes space in your day. It becomes capacity someone else can fill.
An organization may have good reasons to need additional work. What needs to be acknowledged is that more output and a lighter workload are different outcomes. Employees should be able to understand which outcome their organization is pursuing.
If training is introduced with the promise that AI will take pressure off, while performance expectations quietly rise, people have reason to question the promise. They can see that the tool works. They can also see that their day has not become easier.
You are still responsible for what comes out
My work has included training professionals to use software. That background shapes how I look at this conversation. Learning the features of a system is only part of learning how to use it in the work that actually needs to get done.
There is the task demonstrated in training, and then there is the task inside a real organization—with its exceptions, expectations, and consequences.
Consider a trainer using AI to draft instructions. The draft arrives quickly. Someone still needs to check whether the steps match the system people will actually use, whether the language is clear, and whether an important exception has disappeared.
Or consider a manager who receives an impressive summary. Before passing it along, the manager needs to know whether it leaves out the very issue the team must resolve.
The person doing that checking may be the same experienced professional who was told the tool would save time.
From outside, the document looks finished. From inside the work, it still needs attention. That difference matters when someone starts reducing the time allowed for the whole assignment because the first draft arrived faster.
You cannot be expected to produce at the speed of the tool while remaining responsible for every consequence of the result without enough time to review it.
Learning AI is another demand on an already full day
There is also the transition itself.
If an employee’s existing responsibilities remain in place, training adds another commitment. Practice takes time. Learning which tasks the tool handles well takes time. Discovering where it needs correction takes time.
It is easy to tell someone that the investment will pay off later. It is harder to explain where that investment fits between the work due today and the work already overdue.
An experienced worker may be quite willing to learn and still need something taken off the schedule to make learning possible. Willingness does not create additional hours.
The same applies when an organization keeps adding tools. Each may offer a useful feature. The employee still has to understand how it fits, when to use it, and how its output should be checked.
A promise of future efficiency does not erase the workload of getting there.
Sometimes we help the workload grow
The pressure does not always begin with a manager assigning more.
There can be real excitement in discovering what a tool allows you to do. You attempt something that used to require help. You create another version because you can. You start one more task before logging off because it seems so easy to get it moving.
Then the extra effort becomes familiar. What began as experimentation starts to look like your normal capacity.
UC Berkeley Haas researchers observed this pattern during an eight-month study at one U.S. technology company. Employees broadened the work they took on, let work enter former pauses, and kept more tasks moving at once. The findings describe one workplace, but they offer a useful explanation for how enthusiasm can gradually become a heavier working day.[1]
You can enjoy using AI and still need a stopping point. Feeling capable in the moment does not mean the total workload is sustainable.
Why it feels like smoke and mirrors
My concern is the gap between what is promised and what is counted.
The organization sees a faster draft. The employee experiences the checking, the additional requests, the learning, and the pressure to deliver sooner. If only the first part is recognized, the organization and the worker are describing two different versions of the same job.
The efficiency may be real but so is the exhaustion.
This does not require assuming that every employer is deliberately misleading people. It does require asking whether the promise of relief has ever been translated into a decision about workload.
What will be removed? Which deadline will change? Where will practice time come from? Will the recovered time support better work, or will it be filled with more work?
Without answers, “AI will save you time” is an incomplete promise.
The International Labour Organization has also identified work intensification and reduced autonomy among the risks associated with AI at work.[2] The issue deserves attention as part of how work is organized, rather than being left entirely to the employee to manage.
Training needs to be followed by a conversation about the job
For the professional, a useful question after training is: “Now that I can do this differently, what changes about my workload?”
That question is reasonable. It asks the organization to connect the new capability with the conditions under which it expects the work to be done.
A more specific conversation might sound like this: “The first draft is faster, but I still need to review it before I can stand behind it. If we are adding these deliverables, which existing priority moves?”
Employees may not have the power to secure the answer they want. That is precisely why organizations need to take responsibility for asking the question themselves.
Before announcing a productivity success, speak with the people doing the work. Ask whether they have fewer unfinished tasks, adequate review time, and a working day they can reasonably end. Find out what has been added as well as what has been accelerated.
Experienced workers bring judgment that organizations still rely on. That judgment needs room to function. A faster stream of material does not increase someone’s capacity to assess it indefinitely.
If we teach people to use AI and then fill every recovered minute with another expectation, we should be honest about the result: we have increased what they can produce without necessarily reducing what they carry.
The question is whether the person using AI gets any relief—and what the organization is willing to change to make that relief real.