IF AI SAVES TIME, WHO GETS THE TIME BACK? Leaders Decide If AI Builds Capacity or Creates More Work

By Leslie Rohonczy, IMC™, PCC

An employee figures out how to use AI to turn a two-hour daily task into a 20-minute one. On paper, the organization has just gained 100 minutes of capacity every week. In practice, that gain only matters if someone decides what the time is for.

Does it get allocated to deeper thinking, stronger relationships, harder problems? Does the organization reduce staffing? Does the employee just become responsible for producing more, faster? Or does it vanish into the hazy workplace swamp of meetings, messages, and overflowing inboxes?

LinkedIn is filthy with articles either fear-mongering or cheerleading about AI. So many leadership articles seem to focus on how much time AI can save. I'm interested in a different question: what will leaders decide to do with that time?

If every efficiency gain is immediately converted into more volume, we're not improving the system: we're standing in a sinking boat, bucket in hand, while a larger hose pours in work faster than we can bail.

I’m not an AI expert, and I don’t claim to be one. What I do have is a front-row seat to senior leaders trying to make sense of the expectations AI is creating inside their organizations. From that vantage point, recovered time looks less like a technology question and more like a leadership decision.

There's a second question tucked inside this one, and it may matter even more: if AI takes over more of the work people have traditionally used to learn their craft, how will they develop the judgment to know whether that work is any good?

RECOVERED TIME GETS CLAIMED, NOT KEPT

I read about a study done just last year that got me thinking. The Harvard Business School working paper studied 78 employees at IG Group, who were asked to use generative AI to produce online investment articles. Across participants, the average time spent coming up with the concepts for an article fell from about 63 minutes to 23. The writing phase fell from 87 minutes to 22.

Those are significant gains, but they tell us nothing about how the recovered time should be used. Workplace capacity rarely sits around waiting to be thoughtfully allocated. Those gains get claimed, or they evaporate.

In 2025, Microsoft analyzed billions of anonymized Microsoft 365 records and found that the average worker receives 117 emails and 153 Teams messages every weekday. It described an “infinite workday” stretching from early-morning email into late-night inbox checks. The warning leaders should sit with is this: AI can accelerate a broken system unless organizations also rethink the rhythm of work.

I've coached enough senior leaders to know that warning is right on the money. Adding a faster tool to an already-overflowing workday and inefficient systems doesn't create breathing room. It just teaches the broken system to run faster while pouring in just as much, if not more.

THE WORK ISN’T JUST OUTPUT, IT’S DEVELOPMENT

Imagine that you’re a junior analyst who spends hours reviewing data and preparing a first draft, only to notice that two numbers don't quite line up. Much of that labor might be ideal for AI assistance, but it’s not ideal for your development. You’re also learning which anomalies deserve attention, developing a feel for what "normal" looks like, encountering your own assumptions, making mistakes, and discovering why those mistakes matter. You’re building judgment because your apprenticeship is built into your everyday work.

According to a 2026 OECD review, generative AI can be a genuine boost to learning, offering personalized help and fast access to information. But it also flagged a real risk: people leaning on AI output without stopping to evaluate it. Whether that hurts critical thinking long term depends on how the technology gets used, whether it becomes a tutor and practice partner that sharpens our thinking, or something that steals the learning opportunity from us by supplying the answer before we've wrestled with the question. Most organizations haven't yet decided which one they're building toward.

A POLISHED ANSWER CAN HIDE AN EMPTY TOOLBOX

That same Harvard Business School research also compared three groups doing web-analyst work: 12 experienced web analysts, 26 marketing specialists with adjacent skills, and 40 technology specialists whose day-to-day work had little in common with writing investment articles. With AI assistance, all three groups scored about the same on conceptualization. Then came the writing, and the gap showed up: the technology specialists' articles scored roughly 13 percent lower on quality than the other two groups.

Put simply: AI can give you the map, but you still need experience to navigate the terrain. That matters for leaders because AI-assisted work can look impressively complete while hiding how little the person behind it actually understands.

We can no longer assume that a polished first draft reveals much about someone’s real capability. The risk appears when leaders mistake output for judgment. It usually shows up later: when circumstances change, when the AI gets something wrong, or when the employee has to defend the recommendation under pressure. The question is not just “can they produce it?” It is “can they explain it, challenge it, adapt it, and tell you when it should not be trusted?”

RECOVERED CAPACITY NEEDS AN OWNER AND A PURPOSE

Plenty of people have a legitimate claim on recovered capacity. The organization wants output. Employees need relief from unsustainable workloads. Teams need time for learning and collaboration. Customers benefit from more attention. Leaders finally get room for the strategic work that operational urgency keeps crowding out.

So who decides? Leave it undecided, and your existing habits will fill the vacuum for you. Here’s what I’ve noticed: in a volume-driven culture, the time becomes more volume. In a meeting-heavy culture, it becomes more meetings. In an understaffed team, it disappears into the backlog before anyone even notices it existed. "Saving time" isn't a full AI strategy on its own. We still have to decide what the recovered time is for, who benefits, and how we'll explain that decision to our people.

Part of that decision is knowing which friction to protect. Some is worth killing off (nobody needs the character-building experience of reformatting the same information into six different templates). But other effort is developmental, not wasteful. Struggling with a problem before you consult AI builds your reasoning and confidence. Drafting a first recommendation reveals the gaps in your own understanding. Reviewing the source material yourself exposes contradictions that a tidy AI summary would smooth over for readability. Remove the friction, and you create capacity. Remove the developmental effort, and you create dependency.

I'm not suggesting we ban AI from developmental work; that would kill off real learning too. What I am suggesting is designing the process on purpose: have the employee form their own view first, then compare it against the AI's output, identify where they diverge, verify the evidence, and explain the final recommendation in their own words. Sure, this will be slower than just taking the first polished answer, but it builds the judgment they'll actually need when the answer isn't obvious, when the context shifts, or when the technology gets it wrong.

PRODUCTIVITY NEEDS TWO SETS OF METRICS

Most organizations are already tracking the usual productivity measures: cycle time, tickets closed, drafts produced, cost per transaction, AI usage rates, error rates, and throughput per employee. And we can all agree these are important. But they don’t tell us whether AI is building a healthier system or merely helping a strained one move faster.

The missing column should track what actually happens to recovered capacity: hours reinvested in learning, manager coaching hours, rework and escalation rates, employee workload scores, after-hours activity, time spent on complex problem-solving, quality-review outcomes, protected learning time, and whether people can still explain, challenge, and defend the work AI helped produce.

Does the AI technology build up our human capability, or just erode the practice that builds it? If experienced employees developed their judgment by doing the reps, how will less-experienced employees develop theirs without them? Are workloads genuinely getting more sustainable, or just expanding to soak up every bit of gain that AI hands you?

If the only thing people can point to after AI adoption is "we process more work, faster," don't be surprised when employees conclude the reward for becoming more efficient is simply more work. That's not a recipe for enthusiastic adoption. It never has been.

YOUR COACHING CHALLENGE

Over the next two weeks, pick one area of your team's work where AI is already saving time, or soon will. Work through these questions:

What capacity could realistically be recovered, and how would we know?
If nobody makes an explicit decision, where is that time most likely to go by default?
Ideally, what portion should improve performance, and what portion should strengthen learning, relationships, innovation, or workload sustainability?
What judgment have experienced employees built by doing the work AI may now perform?
How will newer employees build that same judgment without the same reps?
Where should AI assist, and where should employees still be expected to think, attempt, question, wrestle with, and decide for themselves?
What would a fair, transparent conversation with employees about recovered capacity actually sound like?

The future of work will be shaped by what AI becomes capable of doing. It will be shaped just as much by what leaders choose to do with the capacity it creates, and by what they still ask human beings to practice, learn, and own.

If this is showing up in your organization after an AI rollout, a retention concern, a succession pipeline where judgment gaps are surfacing earlier than expected, or a reorg redefining what “senior” means, I’d be glad to help you explore it through an executive coaching lens. Reach out for a free exploratory conversation at www.leslierohonczy.com.

For readers who want to explore the research behind this article in more depth, here are the links: