AI Could Save You Three Hours and Still Leave You With More Work
Discover how AI can enhance productivity while increasing workload expectations in your organization.
Article
An operations manager introduces an AI assistant to help her team prepare its weekly performance reports. It's a familiar task that has gradually become more time-consuming as the business has grown. Every Friday, an employee gathers information from several systems, checks the figures, identifies anything unusual and writes a commentary explaining what happened during the week.
The report is useful, but producing it takes several hours that could otherwise be spent helping colleagues resolve operational problems. The first few weeks with AI look promising. The assistant helps organise the information, prepares an initial draft and highlights changes that might deserve attention.
The employee still checks the figures and reviews the conclusions, but the process becomes noticeably faster. Work that previously occupied much of Friday morning can now be completed in a fraction of the time, and the manager is understandably pleased.
A month later, however, something interesting has happened. Because the report is easier to produce, the manager starts asking for a more detailed breakdown of performance. Another department requests a similar report, while senior leadership wants additional commentary about the trends the AI has identified.
The employee is also spending more time checking generated explanations, some of which sound convincing but aren't properly supported by the underlying data.
The original report is still being produced faster than before. But the employee doesn't seem to have gained much time, and the number of requests arriving each week has increased. What began as an efficiency improvement has gradually changed the amount and nature of the work expected from the team.
I think this illustrates an important contradiction in how we discuss AI productivity. We're becoming increasingly capable of measuring how quickly an AI system completes a particular task, but we're not always asking what happens to the time that task previously required.
Does it reduce pressure on employees, allow the business to serve more customers or improve the quality of decisions? Or does it simply create an expectation that people should now produce more, review more and respond more quickly?
I was thinking about this while reading McKinsey Global Institute's September 2026 report, *The AI Economy: Interconnected Forces, Feedback Loops and Speeds of Change*. The report cites US Census Bureau research in which nearly 30% of workers who had used AI at work during the previous week reported saving at least three hours.
Another 31% reported saving between one and two hours. These are self-reported estimates of time saved, rather than independently measured improvements in business performance, but they raise a question that I think deserves much more attention.
## If AI is helping people complete work significantly faster, where is that capacity actually going?
## Saving time on a task isn't the same as improving the business
Our operations employee provides a useful example because the immediate improvement appears straightforward. The weekly report takes less time to prepare, and the employee can complete the first draft with much less manual effort. From the perspective of the individual task, the technology has delivered what the business hoped it would.
The problem is that the report doesn't exist independently of everything else happening in the organisation. The manager uses it to understand operational performance, other departments rely on parts of the information and senior leadership may use its conclusions to make decisions. When producing the report becomes easier, the expectations surrounding it can change.
People may request more frequent updates, more detailed analysis or additional reports that previously seemed too expensive to prepare. The business has effectively lowered the cost of generating information, and that can encourage greater demand for it.
There's nothing inherently wrong with this development. If the extra analysis helps the organisation identify problems earlier, reduce waste or make better decisions, the additional work could be worthwhile. A business that uses AI to provide more valuable information without proportionately increasing its costs may have achieved a meaningful productivity improvement.