Productivity Can Quietly Turn Into Work Intensification
One particularly interesting piece of research complicates the idea that better technology automatically produces more leisure.
A 2025 National Bureau of Economic Research working paper examined AI exposure and how people allocate their time. The researchers found that greater occupational exposure to AI was associated with longer work hours and less leisure, particularly where AI complemented human workers and made them more productive.
That doesn’t mean using ChatGPT for 20 minutes will magically make your workday longer.
It points to a broader economic possibility.
When technology makes each hour of human effort more productive, that hour can become more valuable.
And when an hour becomes more valuable, companies, clients, markets—and sometimes workers themselves—may want more of those hours rather than fewer.
The productivity gain gets converted into additional output.
That is almost the opposite of the future many people imagined.
We pictured machines doing more so humans could do less.
Instead, machines may sometimes allow humans to do more, so we decide to do even more.
AI Creates New Work While Removing Old Work
Another reason the time savings can be hard to see is that technological change rarely consists only of subtraction.
AI removes certain tasks while creating others.
Someone needs to decide which tools to use.
Someone needs to learn them.
Prompts and workflows need to be developed.
Outputs need to be checked.
Information needs to be verified.
Automations need to be maintained.
Employees need to learn new expectations.
Businesses rethink processes.
People experiment with new things they previously could not do.
A 2026 revision of research using Danish labor-market data found that workers reported productivity benefits from AI chatbots, while employers also reorganized work around new AI-related tasks such as content generation, oversight, and AI integration. Average recorded working hours had not meaningfully declined.
This helps explain why someone can truthfully say, “AI saves me a lot of time,” while simultaneously feeling like they have more to manage than before.
Both things can be true.
The Personal Productivity Version of the Same Problem
You don’t need to work for a giant company to experience this.
The same thing happens with personal productivity.
Imagine using AI to plan your week.
You finish the plan in five minutes instead of 30.
Then you ask AI for a workout routine.
Then a meal plan.
Then a better budget.
Then five business ideas.
Then a content calendar.
Then a reading list.
Then a detailed morning routine.
Nothing is necessarily wrong with any of those things.
But suddenly you have seven new systems to maintain.
Productivity tools can make planning so easy that we create more commitments than our actual lives can comfortably support.
AI is particularly good at generating possibilities.
Human beings still have to live them.
That distinction is becoming increasingly important.
Stop Asking Only, “How Much Can AI Help Me Do?”
A better productivity question may be:
What do I want AI to help me stop doing?
That sounds similar, but it changes the goal.
Instead of using every efficiency gain to increase output, deliberately convert some of it into reduced workload.
If AI saves you 30 minutes on routine administrative work, you don’t automatically owe that 30 minutes to another task.
Sometimes the productive choice is finishing earlier.
Or taking a real lunch.
Or thinking without a screen.
Or doing one important piece of work with full attention instead of squeezing three minor tasks into the gap.
Productivity should create capacity.
It shouldn’t automatically consume all of it.
Give Saved Time a Job Before Something Else Takes It
One practical way to prevent the cycle is to decide what happens to efficiency gains before you get them.
For example, you might decide:
- AI-generated time savings before noon go toward focused work.
- Time saved at the end of the day stays saved.
- Automation should remove recurring tasks rather than create room for additional recurring tasks.
- A faster first draft does not mean producing more drafts unless more drafts are actually valuable.
- Finishing the day’s priorities means you are finished, even if technology theoretically makes another hour of output possible.
This creates a boundary around productivity.
Otherwise, any available capacity tends to attract more work.
The to-do list expands to fit the tools.
Use AI to Remove Work, Not Just Accelerate It
There is an important difference between asking:
“How can I do this faster?”
and asking:
“Does this need to be done at all?”
AI is excellent at the first question.
Humans need to remain responsible for the second.
If you receive a weekly report nobody reads, having AI produce it in three minutes is an improvement.
Stopping the report might be a bigger one.
If you attend a meeting that accomplishes nothing, having AI summarize it afterward saves some time.
Eliminating the meeting could save more.
If you publish content merely because you now have tools that make it easy to produce, generating it faster may not improve your results.
Better productivity sometimes means subtraction.
Protect the Human Part of the Work
As AI handles more execution, the scarce resource may increasingly become human attention.
That includes judgment.
Taste.
Prioritization.
Creativity.
Context.
Relationships.
And the ability to decide what is worth doing in the first place.
Microsoft’s 2026 research found that more advanced AI users were increasingly expanding the kinds of work they could accomplish rather than simply completing old tasks faster. Fifty-eight percent of surveyed AI users said they were producing work they could not have produced a year earlier.
That is potentially powerful.
But expanded capability needs boundaries.
Being able to do more things does not mean every possible thing deserves a place in your day.
Maybe the Goal Isn’t Maximum Productivity
For decades, productivity advice has often revolved around optimization.
Wake up earlier.
Plan better.
Batch tasks.
Automate.
Delegate.
Eliminate distractions.
Use better software.
Now add AI to the list.
Many of those techniques genuinely help.
That’s also the idea behind Productivity in 2026: How to Get More Done Without Working More Hours: better productivity should help you accomplish what matters without automatically making the workday longer.
But if every improvement simply raises the amount we expect ourselves to accomplish, maximum productivity becomes a finish line that constantly moves farther away.
Eventually, the better question becomes not:
How much can I get done?
but:
How much actually needs to get done?
There is a point where another completed task contributes less to your life than an hour that remains unscheduled.
There is value in efficiency.
There is also value in enough.
The Real Promise of AI Productivity
AI is still relatively young, and its effect on work will continue changing. Current research already shows both sides of the story: measurable time savings in some settings and evidence that higher productivity does not automatically translate into shorter workdays or more leisure.
That means the technology itself probably will not decide whether AI gives us more free time.
We will.
If every saved minute becomes an invitation to produce something else, AI could make us extraordinarily efficient and leave us wondering why we are still exhausted.
But if we use some of that efficiency to eliminate low-value work, protect attention, shorten repetitive tasks, and deliberately keep a portion of the time we save, the outcome can look very different.
AI can help us do more.
The more interesting opportunity may be learning when not to.
