Generative artificial intelligence promised to return hours to knowledge workers by speeding drafting, summarising and routine problem-solving. Early pilots and academic surveys suggested significant time savings, but companies deploying these tools are now confronting a more complex outcome: faster throughput often feeds a rising tide of additional work rather than a sustained reduction in workload. According to CEO Today, what begins as acceleration can quickly harden into a new baseline of expectation for speed and responsiveness.

Expanding workloads

Evidence from multiple studies shows the phenomenon is widespread. A London School of Economics global survey found employees using AI save an average of 7.5 hours a week, yet observational research at a US technology firm recorded workers using AI taking on a broader mix of tasks, extending work hours and shifting the rhythm of when tasks were advanced. That pattern , where lower friction encourages more frequent task initiation and fewer natural pauses , helps explain why many employees report heavier workloads despite tools that make individual tasks faster. According to Forbes, 77% of employees using AI said their workload increased even as executives remained optimistic about productivity gains.

Expectations vs reality

The mismatch between executive expectation and employee experience is corroborated by industry analysts. Gartner advises CFOs to reset assumptions about how quickly AI investments will convert into broad productivity gains and headcount reductions, noting that only a minority of teams report large productivity uplifts. That suggests firms should be cautious about treating AI speed improvements as automatic capacity release.

The growth of "workslop"

Practical harms from accelerating output without safeguards are already visible. Studies highlighted by ITPro and Axios describe widespread “workslop” , time lost correcting low-quality AI outputs , and associate it with customer complaints, rejections and reputational friction. ITPro estimates US enterprise users spend roughly 4.5 hours weekly fixing AI-generated errors, while Axios-derived research places a measurable monthly cost per worker for repairing substandard AI work. These correction burdens can erode the headline time savings and contribute to fatigue.

Market implications

Broader labour-market analysis paints a nuanced picture of AI’s economic effects. Goldman Sachs projects only a modest net impact on overall employment if current AI use cases expand, while PwC’s 2025 barometer finds AI-skilled workers commanding much higher wages and faster job growth in AI-exposed roles. Together these reports indicate AI can raise individual productivity and pay for some workers even as it reshapes task mixes and concentrates risk in particular occupations.

Rethinking the operating model

For senior leaders the strategic question is shifting from cost removal to operating-model design. The evidence indicates organisations that merely funnel faster outputs into higher throughput risk entrenching cognitive strain and poorer decision quality. CEOs should consider policies that distinguish where speed creates genuine value from where it merely increases churn, and introduce deliberate pauses into high-stakes processes to protect judgement and recovery time.

The upskilling solution

Training and governance are key to converting access into sustainable benefit. The LSE survey noted a large share of employees received no recent AI training, and multiple industry reports stress that poor tooling, inadequate oversight and thin guidance accelerate “workslop” and employee stress. Firms that invest in upskilling, quality-control processes and clear norms around after-hours and cross-functional workload are likeliest to preserve the productivity dividend.

The decisive variable

AI remains a powerful lever, but its net effect depends on how organisations choose to absorb it. Used with disciplined governance, training and a focus on decision quality rather than raw responsiveness, the technology can shift teams toward higher-value work. Left unmanaged, however, generative tools risk becoming an invisible ratchet that expands human load and undermines the very performance they were meant to improve.

Understanding technology is the key to advancing your organisations efficiencies. Click here to visit the Knowledge Hub and keep up to date with the latest news and insights.