By Stuart Kerr, Technology Correspondent, LiveAIWire
The AI cult of productivity promises more output in less time, fewer routine tasks consuming cognitive bandwidth, and better decisions made faster with better information. The productivity gains are real in many documented contexts. What receives considerably less attention is what happens to human experience, creativity, and wellbeing when productivity optimisation becomes the organising principle of working life rather than one dimension of it, and when AI tools designed to maximise output are deployed without consideration of the human costs of the optimisation they enable.
The productivity discourse around AI has absorbed and amplified a broader trend in knowledge work culture that researchers in occupational psychology and public health have been documenting with increasing concern. The combination of always-on connectivity, remote working arrangements that collapse the boundaries between work and rest, and now AI tools that extend the range and pace of producible output are creating working environments that the human nervous system did not evolve to inhabit.
WHO and ILO data on mental health at work estimated that 12 billion working days are lost annually to depression and anxiety, with overwork a primary contributing factor. AI tools that make it possible to work longer and faster do not automatically address this problem, and the AI cult of productivity can intensify it rather than resolve it.
The AI Cult of Productivity’s Optimisation Trap
The optimisation trap in AI-enhanced productivity works as follows. AI tools initially reduce the time required for specific tasks, creating genuine efficiency gains. The freed time is then absorbed by additional tasks, extended working hours, or the expectation of faster turnaround on existing work. The net effect on total workload and total working time is not reduction but intensification: more is done, more is expected, and the benchmark for acceptable output rises with the capabilities of the tools available.
The phenomenon has been described by researchers as the productivity paradox: efficiency gains from technology do not consistently translate into reduced working time or reduced cognitive burden, because the gains are absorbed by expanded expectations rather than converted into rest or discretionary time. Harvard Business Review analysis of AI and cognitive performance found that the most productive use of AI tools in knowledge work requires deliberate slowing down to direct AI outputs thoughtfully, and that organisations deploying AI primarily as a speed accelerator produce faster but not better outputs and report higher rates of worker burnout than those treating AI as a quality enhancement tool.
What the AI Cult of Productivity Misses
The productivity framing of AI leaves out the human activities that produce long-term individual and collective capability but are not measurable as output in any given period. Deep reading, unstructured reflection, sustained attention to a single problem, social connection with colleagues, and the incubation time that creativity requires are all activities whose value is real but whose contribution to productivity metrics is invisible or negative in the short term.
As LiveAIWire’s analysis of how AI is reshaping human cognitive identity and capacity found, the cumulative effect of delegating cognitive work to AI systems is not only about what gets done. It is about what capacities atrophy when they are no longer regularly exercised. This is not a reason to avoid AI productivity tools. It is a reason to use them with explicit intentionality about which capacities they should support and which they should not replace.
What a Better Relationship with AI Productivity Looks Like
The organisations and individuals reported to be using AI most effectively in knowledge work contexts are those that treat AI tools as complements to human judgment rather than substitutes for it, that explicitly protect time for deep work unmediated by AI, and that measure success in terms of output quality and worker wellbeing rather than output volume alone.
Governance frameworks have a limited but real role to play. The EU’s right to disconnect legislation, which gives workers the right to be unreachable outside working hours, provides a template for protecting time from the demands of always-on work environments. As LiveAIWire’s coverage of how AI affects working life differently across demographic groups found, the workers with the least power to negotiate the terms of AI-enhanced productivity expectations are those with the most to lose from the intensification those expectations produce.
The Measurement Problem in the AI Cult of Productivity
One reason the AI cult of productivity persists is that productivity gains from AI are measurable and immediate while the costs of overwork accumulation are delayed and diffuse. An organisation that deploys AI tools and sees output increase in the following quarter has a measurable return on investment. The burnout that follows 18 months of AI-intensified work shows up in healthcare costs, turnover, and reduced creativity that are difficult to attribute specifically to the AI deployment that set the conditions for them.
Several major organisations experimenting with four-day working weeks have found that output quality is maintained or improved while worker wellbeing and retention improve significantly, suggesting that the relationship between hours worked and value produced is substantially less linear than productivity optimisation frameworks assume. As LiveAIWire’s analysis of how AI is helping and hurting workers simultaneously found, the workers with the most accumulated expertise are often the most vulnerable to the productivity intensification that AI enables, because their productivity in AI-augmented environments is high enough to make the intensification invisible until it becomes acute.
What Healthy AI-Assisted Work Actually Looks Like
The organisations resisting the AI cult of productivity are not simply deploying AI tools to maximise output. They are using AI to reduce the proportion of work that is purely mechanical, creating space for the judgment-intensive and creative work that people find meaningful and that produces the non-routine value that AI cannot yet reliably generate.
Practically, this means establishing clear expectations about what AI handles and what people handle, rather than using AI availability as a justification for expanding output expectations without limit. It means monitoring the distribution of freed time across teams to ensure that AI efficiency gains translate into sustainable workloads rather than expanded task lists, and treating burnout data as a lagging indicator of AI deployment decisions that are optimising for the wrong variable, rather than as an HR problem disconnected from technology strategy.
About the Author
Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, cybersecurity, and the social impact of emerging technology. He publishes daily at LiveAIWire.com.