A task that used to take an entire afternoon can now be completed in twenty minutes.
When I speak with senior leaders about this, most see it as a clear improvement. And in many ways, I agree.
Yet I have become increasingly interested in something else.
What happened to the afternoon?
What two studies found
In January, Workday published a survey of 3,200 people who use AI at work. The research was conducted in November 2025.
Eighty-five percent of respondents reported saving between one and seven hours per week through the use of AI tools.
That is a significant amount of time.
What I find more interesting is what happened to those hours.
In approximately one third of the organizations surveyed, the time saved was simply replaced with additional work.
In February, two researchers at Berkeley, Maggie Ye and Aruna Ranganathan, published findings from eight months of research inside a technology company with approximately 200 employees. Everyone in the company had access to AI tools.
The researchers wanted to understand what people were doing with the time these tools were supposed to free up.
What they found was quite different from what one might expect.
Employees were working faster. They were taking on responsibilities that previously belonged to others. Work increasingly extended into periods that had once been used for breaks. And many people were managing several activities simultaneously.
Perhaps the most interesting part was that nobody had explicitly asked them to work this way.
We need to be careful about drawing broad conclusions from this research. It involved one company, approximately 200 employees, and the authors themselves described the work as ongoing. The participants were also not senior leaders.
Still, when I look at these findings alongside the Workday survey, I see an important question emerging.
Saving time does not necessarily mean creating more capacity.
Where the thinking used to happen
One possible interpretation is that we simply need to become more disciplined about how we use the time AI gives back to us.
We save hours, fill them with more work, and then wonder why we do not feel any less busy.
From that perspective, the solution would be to protect the time more deliberately.
I think there is another dimension to this that deserves our attention.
Something I have been noticing throughout this year is how much of our thinking used to happen while we were doing the work itself.
AI is remarkably good at accelerating activities that previously required considerable time.
Writing a first draft. Summarizing a forty-page document. Preparing an analysis that once occupied an entire afternoon.
These activities have become significantly faster.
And yet, the time they required was not necessarily wasted time.
Some of it was where our thinking actually developed.
Certainly not all of it. Still, probably more than we tend to recognize.
Think about preparing an important memo.
When that memo took three hours to write, you were not spending three hours simply putting words on a page.
You were developing your argument as you went along.
Perhaps you reached the second paragraph and realized that your original reasoning was not quite right.
You looked up a figure, considered what it actually meant, and spent a little longer thinking about its implications.
You removed a sentence because, on reflection, it did not accurately represent what you believed.
You may even have abandoned an entire approach halfway through because something became clear to you while writing.
None of this appeared as a separate activity in your calendar.
It was part of the work.
Today, AI can produce a very good memo in twenty minutes.
You read it, evaluate it, perhaps make a few adjustments, and send it.
The result may be excellent.
What may be missing is the process through which you would previously have developed your own position.
Why it does not feel like a loss
This is something I find particularly difficult to recognize, including in myself.
Reviewing and evaluating someone else’s work, or something generated by AI, still requires thinking.
You can read a draft carefully, identify gaps, challenge assumptions, request changes, and go through several iterations.
And quite often, the final result will be better than what you might have produced yourself after a long and demanding week.
So there is no obvious sense of losing something.
Quite the opposite.
You experience yourself as becoming more capable.
You can handle more decisions. The quality of your work remains high. And the speed at which things get done is impressive.
What becomes less visible is the distinction between decisions you have developed through your own thinking and decisions you have reviewed and approved.
Both can feel very similar.
And that is precisely why we may not recognize the difference for quite some time.
What we might notice instead is something more subtle.
Our thinking may gradually become less differentiated.
We develop positions that seem convincing until someone challenges them in depth. Then we discover that we have not really examined the underlying assumptions.
We have opinions on an increasing number of subjects, yet fewer of those opinions have developed into well-founded convictions.
And when we eventually encounter something genuinely new, a complex situation for which no existing analysis or draft provides a useful starting point, we may find that working through it requires more effort than it did two years ago.
One question
How much of your own thinking used to happen while doing work that you now delegate to AI?
I am not referring to how quickly you can read information.
Or how efficiently you can produce documents.
Or how much better prepared you can arrive at a meeting.
I am referring to something more fundamental.
The process of developing a position through thinking, questioning, reconsidering, and sometimes discovering that your original understanding was wrong.
The kind of thinking that takes enough time for your assumptions to become visible before somebody else has to challenge them.
When I raise this question with senior leaders, the initial response is often that nothing has really been lost.
After all, the quality of the output has improved.
Yet when we explore the question a little further, that initial certainty often begins to change.
Of course, there are many situations where speed is exactly what we need.
Most of the information and decisions that cross a senior leader’s desk do not require three hours of reflection. They never did.
The point is not that we should return to slower ways of working simply because they are slower.
It is about recognizing what may have been happening during those hours, beyond producing the final result.
And understanding whether we have created another place for that thinking to happen.
Because perhaps the most important question is no longer how much more work we can get done.
It is where our own thinking happens when the work that previously supported it no longer requires us to do it.
If you recognize this pattern across an entire leadership team, rather than just in your own working week, an organizational briefing is often a good starting point for exploring it together.
Request an organizational briefing →
Sources
Workday. Beyond Productivity: Measuring the Real Value of AI. Released 14 January 2026. Survey of 3,200 full-time employees who actively use AI, working in organizations with annual revenue exceeding $100 million across North America, EMEA, and APAC. Research conducted by Hanover Research in November 2025.
newsroom.workday.com
Ye, X. M., & Ranganathan, A. AI Doesn’t Reduce Work—It Intensifies It. Harvard Business Review, February 2026.
hbr.org
University of California. AI promised to free up workers’ time. UC Berkeley Haas researchers found the opposite. 26 February 2026. Methodological details for the research above, including an eight-month ethnographic study in a 200-person US technology company and more than 40 semi-structured interviews. The authors describe the research as ongoing.
universityofcalifornia.edu
