Learning investment emerges as AI’s key differentiator

A significant share of the productivity gains promised by workplace AI is being eroded by rework: employees are spending time correcting errors, rewriting content and verifying outputs, according to new research.

Rather than pointing to a failure of adoption, the findings suggest that organisations are struggling to convert speed into value because workforce skills, learning support and job design have not kept pace with AI deployment.

The study, Beyond productivity: Measuring the real value of AI, finds that while most employees report saving time through AI, nearly 40% of those gains are offset by low-quality output that requires human intervention. Only 14% of employees consistently achieve what the researchers describe as ‘net-positive’ outcomes from AI use.

The research, based on a survey of 3,200 employees and leaders across North America, EMEA and Asia-Pacific, highlights what it calls an ‘AI productivity paradox’. Although 85% of employees say AI saves them between one and seven hours a week, much of that time is absorbed by verification and correction work.

The research suggests that the difference between organisations that realise value from AI and those that do not is less about technology choice and more about how time savings are reinvested.

‘For every ten hours of efficiency gained through AI, nearly four hours are lost to fixing its output,’ the report says, describing this hidden loss as an ‘AI tax on productivity’.

The burden is not evenly distributed. Employees who use AI most frequently are also the most likely to experience rework. Daily users are overwhelmingly optimistic about AI’s potential, with more than 90% believing it will help them succeed in their roles, but 77% say they review AI-generated work as carefully as, or more carefully than, work produced by humans.

• Nearly 40% of AI time savings are lost to rework such as correcting errors and verifying output
• Only 14% of employees consistently achieve clear, positive net outcomes from AI use
• 85% say AI saves them between one and seven hours a week, but much of their gains are absorbed by rework
• 77% say they review AI-generated work as carefully as, or more carefully than, human output
• While most leaders cite training as a priority, only 37% of those facing the highest rework have increased access to training

Younger employees appear to be carrying a disproportionate share of this workload. Employees aged 25 to 34 account for nearly half of those experiencing the highest levels of AI-related rework, despite often being perceived as the most digitally confident.

The research links this pattern directly to learning and capability gaps. While two-thirds of leaders say skills training is a top priority, only 37% of employees dealing with the highest levels of AI rework report having increased access to it.

‘Employees are using 2025 tools inside 2015 job structures,’ the report says, noting that in almost nine in ten organisations fewer than half of roles have been updated to reflect AI capabilities. As a result, employees are expected to deliver higher-quality outcomes with AI without clear guidance on where automation ends and human judgement begins.

This lack of role clarity and skills support is particularly acute in functions such as HR, where work is judgement-heavy and quality thresholds are high. In these environments, ‘good enough’ output is rarely acceptable, leaving employees to absorb the time cost of validation and correction themselves.

By contrast, organisations that see stronger returns from AI appear to be making more deliberate learning and workforce design choices. Employees who report positive AI outcomes are significantly more likely to say their organisation has reinvested time savings into skills development, collaboration and higher-value work, rather than simply increasing workload.

• 79% of employees with positive AI outcomes report increased access to skills training, compared with far lower levels among high-rework users
• 57% of these employees say their organisation has increased investment in collaboration and team connection, indicating time savings are being reinvested in higher-value work
• Employees achieving net-positive AI outcomes are twice as likely to receive substantial skills training as those struggling to realise value
• Among high-usage but low-return AI users, only 37% report increased access to skills training, despite heavy reliance on AI tools
• At organisational level, 39% of AI savings are reinvested in technology compared with just 30% in workforce development, despite stronger outcomes where people investment is higher

‘The organisations realising the greatest value from AI treat saved time as a strategic resource,’ the report says. ‘They reinvest in upskilling their teams, improving collaboration and strengthening judgement-driven work.’

The study was conducted by Workday. Gerrit Kazmaier, president, product and technology at Workday, said the findings show that many organisations are leaving employees to manage quality risks individually:

‘Too many AI tools push the hard questions of trust, accuracy and repeatability back onto users.’

‘AI should do the complex work under the hood so people can focus on judgement, creativity and connection.’

The research concludes that measuring AI success purely in terms of hours saved risks overstating its impact. Instead, it argues, organisations need to assess net value, accounting for time lost to rework, and align investment in technology with sustained investment in learning, skills development and role redesign if AI is to improve outcomes rather than simply accelerate activity.

Download the report
Beyond productivity: Measuring the real value of AI