AI isn’t reducing workload. It’s quietly expanding it.

This keeps coming up in conversations. Leaders expect AI to free up time. Teams report being faster. Yet nobody feels less busy.

A recent Harvard Business Review article from February puts some structure around why.

In an eight-month study of a mid-sized technology company, generative AI didn’t shrink work. It intensified it. Employees worked at a faster pace, took on broader responsibilities, and stayed engaged for more hours of the day. None of this was mandated. It happened because AI made extra work feel easy to start and oddly satisfying to complete.

Three patterns stood out.

First, task expansion. Product managers, designers, and researchers stepped into engineering-adjacent work that previously required extra support or headcount. That work didn’t disappear. It shifted, often landing back with senior engineers who had to review, correct, and guide AI-assisted output.

Second, blurred boundaries. AI removed the friction of starting. Small prompts crept into lunch breaks, meetings, and evenings. The workday didn’t feel longer, but it stopped ending cleanly.

Third, constant parallelism. People ran multiple AI threads at once. It created momentum, but also more checking, more context switching, and more cognitive load than before.

From a hiring perspective, this matters.

AI rarely removes the need for people. It redistributes pressure. Teams assuming AI reduces staffing needs are often underestimating review, decision, and integration work. Senior engineers become bottlenecks. Junior staff look productive faster but still rely on experienced judgment. Contractors are used to absorb overflow rather than clearly defined delivery.

AI doesn’t automatically create capacity. It changes where pressure sits. Short-term output improves. Medium-term strain often follows.

The organisations navigating this best aren’t trying to slow AI down. They’re paying closer attention to where scope expands, where review effort accumulates, and where intensity replaces capacity.

The question isn’t whether AI boosts productivity. It’s whether that productivity is being converted into sustainable work.