The executive consensus on AI is settled. 96% of C-suite leaders expect the tools to boost productivity. In New Zealand, 93% of firms report improved worker output. The numbers look like vindication.
Ask the people actually using the tools and the story inverts. The Upwork Research Institute’s 2024 survey of 2,500 executives and employees found 77% of workers using AI said it added to their workload, not reduced it. Worse, 47% said they had no idea how to achieve the productivity gains their employers expected. That is not a technology failure. That is a management failure.
Denser days, longer weekends, less focus
ActivTrak analysed 443 million hours of digital activity across 1,111 organisations between 2023 and 2025. The findings demolish the idea that AI lightens the load. After adoption, email activity rose 104%, chat and messaging jumped 145%, and business management tool use climbed 94%. The average weekday shrank by a trivial nine minutes, but weekend work surged over 40%, with Saturday productive hours up 46% and Sunday up 58%. Focus time fell to a three-year low, down 23 minutes daily for AI users.
A UC Berkeley eight-month study at a 200-person tech company tracked the same pattern. Workers broadened their scope, product managers started writing code, researchers absorbed engineering tasks. After an initial productivity surge, they reported more intense workdays and less work-life balance. AI did not replace work. It compressed it.
The rework tax nobody budgeted for
Workday’s global study of 3,200 employees and leaders identified what may be the most expensive blind spot in the AI rollout. Organisations are losing nearly 40% of expected productivity gains to employees fixing low-quality AI outputs. Highly engaged workers spend approximately 1.5 weeks per year correcting AI-generated content. Only 14% of employees consistently achieve net-positive outcomes from AI use.
Those 93% of New Zealand firms reporting gains are almost certainly measuring input, not output.
Three tools is the limit, most firms deploy seven
Boston Consulting Group surveyed 1,488 workers and coined the term “AI brain fry” for the mental fatigue from constant AI interaction. 14% of AI-using workers reported the condition. Those affected showed 33% more decision fatigue, 11% more minor errors, and 39% more major errors. A senior engineering manager described the experience: “I kept bouncing between them, double-checking every little thing. But instead of moving faster, my brain just started to feel cluttered.”
Harvard Business Review research pinpointed a critical threshold at three concurrent AI tools, beyond which productivity gains plateau and errors climb. The average organisation now deploys seven or more. Almost every company scaling AI is operating well past the point where the research says it helps.
New Zealand’s particular vulnerability
Professor Jarrod Haar’s research found that high AI users in New Zealand are 2.9 times more likely to experience severe job burnout. Unlike traditional software where updates are incremental, AI tools evolve constantly, demanding continuous adaptation.
The governance gap compounds the problem. 84% of NZ knowledge workers are already using generative AI, but only 47% of employers actively encourage it. That means the majority of workers are navigating AI adoption without formal training, governance, or workload adjustment. Victoria University of Wellington researcher Talitakuum Ekandjo found workers describing AI assistants as oversimplifying the messy, collaborative nature of real work. Mood-tracking features in productivity tools made workers “always suspicious” about who was reading their data.
The sweet spot exists but almost nobody is in it
ActivTrak’s data contains one genuinely useful finding. Employees spending 7-10% of their work hours in AI tools showed the highest productivity at 95%. The problem is only 3% of users fall within that range.
The cost of ignoring this is concrete. WorkTime’s 2026 analysis puts burnout costs at $10,824 per manager per year, with 89% of those costs from presenteeism. Burned-out employees are nearly three times more likely to leave. Disengagement risk has risen to 23% across AI-adopting organisations, up from 19%.
For New Zealand business leaders, the question is no longer whether to adopt AI. It is whether you have redesigned jobs around the tools or simply piled them on top of existing workloads. The evidence, overwhelmingly, suggests the latter. And the bill for that choice is already arriving in weekend hours, rework cycles, and resignations.
Sources
- HCAMag NZ: Kiwi workers are using AI they don’t trust (2025)
- IT Brief NZ: AI speeds up workdays but doesn’t lighten workloads (2025)
- iStart NZ: AI’s hidden catch – Productivity with a cost (2025)
- HCAMag NZ: AI productivity gains offset by rework costs, study finds (2025)
- HCAMag NZ: Too much AI causing ‘brain fry’ among employees (2025)
- Harvard Business Review: When Using AI Leads to Brain Fry (2026-03)
- MindMatters Clinic: AI Burnout and Workplace Wellbeing in NZ (2025)
- RNZ: Beyond the hype – What workers really think about workplace AI assistants (2025)