In the 1980s, economist Robert Solow famously observed that computers were everywhere, except in the productivity statistics. Decades later, as Artificial Intelligence (AI) rapidly embeds itself into the modern workplace, a familiar paradox is re-emerging.
AI adoption is accelerating across organisations, promising speed, scale and efficiency. Across ASEAN, organisations are navigating a new phase of transformation shaped not only by AI, but also by geopolitical complexity and intensifying competition for digital and AI talent. The shift is no longer just about digitisation or cost efficiency but about intelligence it can provide its users.
Yet beneath these gains lies a quieter, often overlooked cost in the form of a growing “productivity tax.” Employees are spending increasing amounts of time validating, correcting and rewriting AI-generated outputs before they can be used.
Workday’s recent research highlights the extent of this shift. In Asia Pacific, half of workers are spending at least an hour each week clarifying or correcting AI-generated content.
Notably, this burden is not evenly distributed. Managers, directors, and those responsible for oversight and decision-making, are most likely to shoulder the task of validating and refining AI-assisted work. Their roles are evolving into that of quality gatekeepers, ensuring outputs meet organisational and contextual standards.
The result is a modern version of Solow’s paradox - more automation, but not less work. Today, the constraints on productivity have not disappeared; they have simply shifted.
The hidden causes of the productivity dip
AI promised to give us back time - and on the surface, it has. They generate drafts, summarise information and automate routine tasks in seconds. Based on our research, employees are saving approximately 1-7 hours a week using AI. But time saved isn’t necessarily translating into business value. Much of the time saved upfront is redirected towards reviewing, editing and validating outputs, diminishing the net productivity gains.
This is largely because AI-generated content often lacks the nuance, context and accuracy required for real-world application. Employees must verify facts, adjust tone and correct inconsistencies before outputs are usable.
In many cases, uncertainty around the reliability of AI further compounds the issue. Without clear guidelines or confidence in the tools, employees tend to over-check, rework and second-guess outputs, eroding efficiency gains.
The challenge becomes even more pronounced in high-stakes domains such as hiring, legal and finance. In these areas, errors carry significantly greater consequences, from compliance breaches to financial loss and reputational damage. As a result, AI-generated outputs are subject to rigorous human scrutiny rather than being acted on directly.
AI systems, while powerful, often lack full visibility into organisational context, regulatory nuances and edge cases. This makes their outputs less dependable in complex or sensitive scenarios.
Ultimately, while AI accelerates the first draft of work, it rarely accelerates the final version. The bottleneck hasn’t disappeared; it has merely shifted.
The reinvestment imperative
Despite these challenges, AI still holds significant potential to drive efficiency. Addressing this paradox requires a shift in how organisations think about AI, from tools that assist work to systems that can responsibly perform work.
The real value of AI emerges when organisations deliberately reinvest time savings into higher-value activities such as strategic thinking, innovation and employee development. Without reinvestment, efficiency gains are simply neutralised.
- Jess O’Reilly, General Manager, ASEAN, Workday.
The time saved is often absorbed by low-value tasks, additional layers of review or increased workloads, leaving overall productivity unchanged.
To address this, organisations must move beyond ad hoc AI adoption and establish clear guardrails. This includes defining which tasks are appropriate for AI, setting expectations for verification and identifying where human sign-off is required. Clear frameworks help reduce unnecessary rework while maintaining quality and accountability.
Equally important is practical AI upskilling. Rather than focusing on abstract theory, organisations should equip employees with hands-on skills: how to craft effective prompts, recognise strong versus weak outputs, and quickly validate accuracy and relevance. When employees know how to work with AI effectively, they spend less time correcting it.
Going beyond productivity: reshaping the employee experience
AI’s impact is not limited to productivity metrics. It is also reshaping how work is experienced.
On one hand, faster task completion can create implicit pressure to take on more work within the same timeframe. This introduces the risk of work intensification, where employees produce more output without a corresponding reduction in workload. Over time, this can offset the perceived benefits of AI.
On the other hand, AI meaningfully reduces the burden of repetitive, manual and administrative tasks. This allows employees to focus on more meaningful, higher-value work, lowering day-to-day cognitive load.
Encouragingly, employees in the region report relatively high levels of trust in their managers to allocate workloads appropriately. This trust plays a critical role in ensuring that AI-driven efficiency gains do not automatically translate into unsustainable expectations.
There are also early signs of positive well-being outcomes. Respondents in the region have reported reduced stress levels and a lower risk of burnout following the adoption of AI tools. This suggests that, when implemented thoughtfully, AI can improve not just how much work gets done, but how it feels to do it.
Turning AI speed into real value
Strategic deployers of AI will be those that treat AI not merely as a chatbot, but as a superintelligent system capable of operating as a true collaborative partner. In practice, this means deploying AI deeply embedded in their business context, their people, and their financial realities, so it can act with real autonomy within the right guardrails.
The re-emergence of a productivity paradox in the age of AI highlights a crucial truth: technology alone is not enough. Faster tools do not automatically create better outcomes. It’s also about using them better.
Jess O’Reilly is General Manager, ASEAN, Workday.
