The AI Dividend: Why AI Is Making Work More Intense, and How Teams Keep the Time It Saves

AI was supposed to give us time back. It has for some teams. What they do with it is the question nobody is asking.

The early research is uncomfortable to hear. The time AI gives us? Most teams aren’t using it for more thinking, rest or connection. They’re using it to get more work done. Not because anyone asked for it, mind you. Just because they can, with the help of AI.

We call this the AI dividend. Every team gets it. Very few make a decision about what to do with it.

What is AI work intensification?

AI work intensification is when AI increases the speed, breadth and hours of work faster than it reduces the effort required, so people end up doing more work overall. It’s usually a slow and voluntary process, so it’s easy to overlook until you start noticing people getting tired more easily, making worse decisions and so on.

It’s also known as workload creep or AI burnout.

What the research shows

The clearest evidence to date comes from an eight-month field study by UC Berkeley researchers Aruna Ranganathan and Xingqi Maggie Ye, published in Harvard Business Review in February 2026. They studied a US tech company of around 200 people where the use of AI was encouraged but not required.

Key takeaways from the study:

• People worked more quickly, took on a wider variety of tasks and expanded the number of hours they spent on work, without being prompted to do so.
• Because AI made tasks so easy to start, work crept into breaks, early mornings and evenings.
• The researchers note that this can result in cognitive fatigue, burnout and poorer decision-making. They also point out that the short-term gains from AI can quickly be undone by declining quality and rising turnover.

Other 2026 findings echo these conclusions. A report summarised by HRD Connect describes a new form of mental exhaustion that comes from overseeing, reviewing and managing AI outputs. Workplace metrics show people handling more tasks and multitasking more, while the amount of deep, focused work declined after AI was introduced.

None of this means AI is bad for us. Microsoft’s 2026 Work Trend Index found that most people who use AI say it gives them more time for high-value tasks, and that they’re producing things they couldn’t have a year ago. It also found that the organisation around a person (its culture, managers and norms) matters nearly twice as much as individual skill in whether AI delivers value.

See what I mean? The dividend is there. The organisation decides what to do with it.

Why AI intensifies work: three mechanisms

Task expansion

When a draft email, a summary or a first version of code can be produced in minutes rather than hours, you say yes to more. You take on more work. Work that other roles used to do. Side projects. A broader definition of “done.” Every yes is justified. Collectively, you’re filling your weeks with yeses.

Blurred boundaries

No more blank pages. With AI, the natural brakes are off. It’s easy to “just prompt something” during lunch, on the couch or before your first cup of coffee. The natural breaks in your day are quietly dissolving.

Invisible supervision

Working with AI isn’t a passive experience. Someone has to review outputs, catch subtle mistakes, toggle between tools and hold multiple half-finished threads in their head. It’s real thinking time. And it’s not showing up in anyone’s timesheets or workload planning.

The common thread: the extra work is mostly self-created. Well-meaning, too. Which is why policies and rules won’t solve this. This is a question of team norms and capacity.

The AI dividend: where should the saved time go?

If AI is saving your team five hours a week, there are only four places that time can go:

Most teams default to the first row without ever choosing it. Smart teams split the dividend across all four.

A human AI practice: five team norms

The Berkeley researchers recommend that companies create an “AI practice”: a set of team norms that includes intentional pauses, thoughtful task sequencing and keeping people grounded in human connection. Here’s what that looks like in a typical week in the life of a team.

1. Name the dividend. Once a month, ask your team where the time AI saved you actually went. Then decide where you want to put it next month.
2. Add a decision pause. Before any AI-assisted output goes to a client, to production or to leadership, take two minutes to step away and double-check your assumptions. Faster drafts need slower checks.
3. Sequence, don’t stack. Set a limit on the number of AI-assisted workstreams one person juggles at once. Finish, then start.
4. Breaks are breaks. Lunch and the end of the day are not “just a moment to prompt something.” Agree on that as a team.
5. Keep a human sync. Hold one team conversation each week that’s about people, not work: What’s the pace like? What’s draining? What’s working? Keep it short, honest and screen-free.

You don’t need software for any of these. You need a team that has the energy and trust to have these conversations.

Want to start these conversations with your team in the room together? Check out our team formats →

What HR and team leads can do this quarter

For People & Culture teams

• Add one question to your next pulse survey: “Since we started using AI tools, has your workload gone up, down or stayed the same?” The answer tells you whether the dividend is being put to use or absorbed.
• Make AI rollout and wellbeing part of the same programme. Every AI training can end with ten minutes of conversation about pace, boundaries and recovery.
• Start with managers. They set the norms, and they’re often the most stretched people in the room. In Germany, Gallup research finds that only a small minority of managers feel supported in their role (Gallup).

For team leads

• Pick two of the five team norms above and run them for four weeks.
• Model the boundary yourself. If you prompt at 22:00, your team will read that as the expectation.
• Watch for the subtle signs of fatigue: shorter tempers in reviews, more small mistakes, people skipping breaks they used to take.

For founders and leadership

• Decide what the AI dividend is for, and say it out loud. “We expect AI to give us more output” and “we expect AI to give us better decisions and a sustainable pace” lead to very different companies.

Where ARUA fits

We work with teams on the human side of fast, AI-enabled work: attention, recovery, emotional regulation and connection.

Check out our team formats → · Reach out to us about your team →