In February 2026, a paper was published in Harvard Business Review by Aruna Ranganathan and Xingqui Maggie Ye that described the results of an eight-month study of how generative AI changed work habits at a mid-sized U.S.-based technology firm. They found that AI, while making tasks go quicker and easier, allowed employees to not just work faster, but ironically work longer hours and take on more tasks outside of their role. Not because they were asked to but because AI made work easier, and that efficiency brought momentum that was easy to continue.
If an hour task can be automated and completed in 10 minutes with AI, we don't take 50 minutes off. We us AI to do another 10-minute task, and another, and another to fill the hour. This rapid task completion is driven by dopamine and satisfied by serotonin: event, action, result—which is the same reason people doom scroll on social media. Being able to do satisfying tasks easy is the quickest way to sustain focus and attention.
That feeling doesn't tell you to stop. It tells you to keep going. Imagine if you were in an airport walking on the people mover to increase the speed of your walking. You would feel energized that it was easier to travel farther, so you likely would walk instead of taking the railway between terminals.
This concept that efficiency breeds demand, not conservation, is called Jevons paradox. When technology makes the use of a resource more efficient, humans don't do less work, they do more. William Stanley Jevons made this observation in his 1865 book, The Coal Question. At the time, Britain was burning a lot of coal, and the hope was the use of the steam engine would reduce their coal use. And it was true, steam engines helped coal-powered machines burn less coal. But, because they were cheaper to run, the factories used them more.
Jevons' insights went forgotten for 100 years until the 1970s U.S. energy crisis, when researchers found him again because they saw cars becoming more fuel efficient, making driving cheaper and leading to people driving more and therefore using more fuel.
It came back again in the 1980s when computers made office work more capable. We didn't reduce our work; we expanded what we could do.
I'm old enough to remember in the 1980s when working as an intern in an office, I would write out a report then give it to a team in another room on Wang terminals who would type it up in the evening and leave a printout on my desk in the morning. When we started using individual word processors (WordStar was my first, but WordPerfect was my favorite), that reduced production time and potential iterations significantly. Instead of handwriting a report in a day, giving it to a typist, and getting the output the next day, I could write, edit, and produce at the same time and create multiple reports a day.
Here are other examples from other industries:
Accountants using spreadsheets made accountants more useful, speeding up simple math and able to do more complex equations easily, thus expanding the types and size of projects they can do. And equations become sharable across the organization, instead of gatekept by someone in their personal notebook.
Architects using AutoCAD: Twenty years ago, I had a conversation on a train with a veteran architect and asked him about the transition from drafting to computer-generated design. He said when they started using computers, clients asked to produce more iterations, the firm used smaller teams to tackle bigger projects, and this ultimately increased workload.
Digital cameras make unlimited shots cheap, so professional photographers now shoot hundreds per session, spend hours editing, and are expected to deliver more options. In the old days, you shot 36 pictures, developed them, and the customer just picked the best of the lot.
It turns out, humans don't want a certain amount of convenience or capability. We want more, and efficiency is how we get it. When friction is removed, work is expanded to fill the gap. Jevons paradox is not a human flaw in how we respond to AI; it's a feature of human nature. We will always fill available capacity. We will always expand to meet new capability.
Which brings an underrated risk of AI by making tasks easier to do by yourself. You can now accomplish an enormous amount entirely alone: research, write, plan, build—all without any other human assistance. But with it, we also lose connection with people, and a certain kind of thinking and collaboration that only happens between people.
The best use of time AI gives back shouldn't be to do more tasks. It should be more time with other humans. Conversations that don't have an agenda. Perspectives that aren't curated by a model trained on your own preferences. Disagreements that are uncomfortable and therefore useful.
In an era when AI can help you do more of the work, the human parts of work—such as relationship building, trust, disagreement, even surprise—become more valuable for our connection and enrichment.
Throughout human history, each new technology advancement promised to give people time back, and each one found that time immediately claimed by something else.
[RELATED: 2026 SecureWorld Theme: Timeless Cybersecurity]
Whether you're the one deciding what fills the time freed up by AI, or that gets decided for you, the workers who navigate the AI era without over scheduling themselves and burning out won't be the ones who automated the most. The people who succeed will be the ones who decided what they are working for. Purpose is the only thing that keeps the machine from running you.
Use this recaptured time toward the work that AI can't do; judgment, trust-building, creativity, and the kind of slow thinking that produces strategy rather than output.
But this vision also requires leaders who are willing to help their staff protect this space, not just assign more tasks to fill it. This is a conversation leaders will need to figure out, and learn how to accommodate.
[REALTED: Leadership in the Age of AI]
