AI productivity gains undermined by rising rework burden

Workday has published research linking AI-driven productivity gains to higher levels of rework, as employees spend time checking and correcting low-quality outputs instead of using the time saved for higher-value work.

The study found that many organisations report clear time savings from AI tools, but those gains do not always translate into better outcomes. Gaps in training, job design, and day-to-day processes can limit returns from AI adoption.

Australian picture

In Australia, the findings point to mixed progress. Only 21% of employees said they use AI tools daily, while 51% use them a few times a week. Even so, 75% said AI has made them more productive over the past year.

The survey suggests the time saved can be significant. Nearly half of Australian respondents (47%) said they save four to seven hours a week using AI. That level of claimed time saving has made AI a regular feature of work for many, even if daily use remains limited.

Quality remains a constraint. More than a third of Australian employees (37%) said they spend one to two hours a week correcting or rewriting low-quality AI outputs. This points to a recurring overhead, particularly where work must meet internal standards or be used in customer-facing contexts.

Jo-Anne Ruhl, Vice President and Managing Director, Workday Australia and New Zealand, said the issue is less about access to tools and more about operational integration.

"Australian organisations are starting to see real productivity benefits from AI, but many are still working out how to turn that into lasting value. Our research shows employees are saving time, yet a significant portion of that is being spent checking and correcting AI outputs. That tells us the challenge isn't access to AI, but how well it's embedded into everyday work. The next step is helping people build the skills, confidence and systems they need to use AI in a way that supports high-quality outputs and good judgement, rather than more rework," said Ruhl.

The Australian results also suggest a disconnect between leadership expectations and employee experience of training. While 86% of leaders said AI time savings are being reinvested into skills and learning, only 52% of employees reported seeing increased training opportunities.

Global findings

Globally, 85% of employees said they save between one and seven hours per week using AI. Workday argued that much of this time is consumed by rework, including correcting errors, rewriting content, and verifying outputs from generic AI tools.

The study estimated that nearly 40% of AI time savings are lost to rework. It also found that only 14% of employees consistently see clear, positive net outcomes from AI use, which it attributed to organisations not updating roles and processes to match new tools.

Frequent AI users were optimistic but reported a heavier verification burden. Among employees using AI every day, more than 90% believed AI will help them succeed. At the same time, 77% of daily users said they review AI-generated work as carefully as, or more carefully than, work completed by humans.

The report also pointed to uneven impacts across age groups. Employees aged 25 to 34 accounted for 46% of those experiencing the highest levels of AI rework, challenging assumptions that younger, more digitally confident staff will automatically see smoother adoption.

Training was another area of risk. While 66% of leaders cited skills training as a priority, only 37% of employees facing the most rework said they have access to training.

Work design also emerged as a limiting factor. In 89% of organisations surveyed, fewer than half of roles had been updated to reflect AI. This suggests many staff are applying AI tools within job structures designed for earlier workflows, with limited changes to responsibilities, controls, or hand-offs between teams.

Where savings go

The research also tracked what organisations do with the time AI saves. It found organisations were more likely to reinvest those savings into technology (39%) than employee development (30%). Another 32% said time savings lead to increased workload rather than structured reinvestment.

Workday said organisations that see stronger returns treat saved time as a resource that requires planning, and highlighted links between positive outcomes and increased training. Employees reporting positive AI outcomes were more likely to use saved time for deeper analysis, better decision-making, and strategic thinking (57%) rather than taking on more tasks. They were also more likely to report increased skills training (79%).

Gerrit Kazmaier, President, Product and Technology, Workday, said the burden of trust and verification often falls to individual workers.

"Too many AI tools push the hard questions of trust, accuracy, and repeatability back onto individual users," said Kazmaier.

The report was based on a survey of 3,200 full-time employees across North America, Asia-Pacific, and Europe, the Middle East and Africa. Respondents worked at organisations with more than USD $100 million in annual revenue and were active users of AI technology.