by Mick Lavin, Coach, Agile Coach, Mentor
Here’s a scenario that’s playing out in organisations across every sector right now. A team adopts several AI tools. The initial time savings are good, documents get drafted faster, data gets summarised more quickly, meeting notes appear automatically. Productivity looks up.
Six months later, that same team is exhausted. Output volume has increased. Quality has become inconsistent. Employees are spending hours correcting AI-generated content they didn’t ask to review. Burnout rates are climbing. And nobody is quite sure why.
This is the AI Productivity Paradox, one of the most underappreciated challenges facing organisations in 2026. The tools that were supposed to free up time are, in many companies, consuming more of it. Understanding why is the first step to doing something about it.
The 37% Problem: “Workslop” and the Hidden Tax on Your Team
Research by Workday in early 2026 found that while generative AI saves knowledge workers between one and seven hours per week on routine tasks, a staggering 37% of that saved time is immediately lost to remediation and cleanup.
The culprit is what researchers have labelled workslop, AI-generated output that appears polished and professional on the surface but ultimately creates a massive hidden drain on productivity through rework, verification, and correction.
It shows-up in three main ways:
Content Slop: Text that sounds credible but is factually incorrect (hallucinated), generically written, or strategically empty. Someone in your team still has to read it, fact-check it, rewrite it to match your brand voice, and fix the fabricated statistics before it goes any further.
Decision Slop: AI generates fifty marketing taglines or thirty policy options in three seconds. Without clear criteria for evaluation, this creates analysis paralysis, longer meetings, more cognitive energy spent filtering mediocre options, and slower decisions. There are ways to mitigate some of this, but people need education in best practices to make this useful.
Process Slop: AI tools that don’t integrate cleanly into existing systems and force employees to manually copy outputs between platforms, reformat content, and bridge incompatible workflows. This invisible “shadow work” is a significant and largely unmeasured productivity drain. For anyone who has tried to use AI to create a presentation, you will understand the issue.
And while many people now use AI to enhance their writing, make those difficult emails feel less hard-hitting, or even auto-respond to email, we can feel that the text has been touched by AI. This may be fine, when we can feel the effort was to clarify and enhance understanding. However, you may have experienced those emails that are wordy, overly sycophantic, and really says nothing! We spend our time reading and re-reading the email to figure out what the sender meant. We may need to respond to ask for clarity. We waste our time trying to understand someone else’s “Content Slop”. This is where the 37% of remediation and cleanup is spent and unfortunately, this is the time someone else saved being passed on to us.
“AI Brain Fry” – The Real Cost of Managing Intelligent Systems
Beyond workslop, there’s a deeper cognitive challenge emerging in AI-intensive workplaces.
Employees are increasingly expected to manage, direct, and oversee multiple AI systems simultaneously, each with different interfaces, different capabilities, and different failure modes. Unlike managing human colleagues, who naturally hold context and develop shared intuition over time, AI agents are fundamentally amnesiac. They require constant re-prompting, context-setting, and verification. Much of the time spent providing the context could have been spent drafting the content.
The mental strain of acting as the sole coherent thread connecting multiple independent algorithmic tasks is exhausting knowledge workers. Researchers and employees themselves have begun using the term “AI brain fry” to describe this state of acute cognitive fatigue.
The numbers support the experience. A comprehensive study by the Upwork Research Institute found that the most productive AI users within organisations were significantly more likely to report severe burnout, disengagement, and an intention to quit. The time saved by automation didn’t translate into recovery time. It translated into higher executive expectations, expanded scope, and relentless “always-on” intensity.
Context-switching, quickly moving from one task to another without adequate breaks, and being alert to the need to validate AI outcomes, are causing stress and burnout at an alarming rate.
This has direct implications for HR and legal professionals. Increased burnout claims, stress-related absence, and constructive dismissal risks don’t disappear because they’re AI-related in origin. They land on your desk.
The Jevons Paradox in Modern Workplaces
There’s a 19th-century economic principle that maps perfectly onto this problem. The Jevons Paradox describes how, as a resource becomes more efficient to use, overall consumption of that resource tends to increase, not decrease.
In knowledge work, this plays out precisely. Improving the efficiency of drafting emails doesn’t lead to fewer emails. It leads to more emails. Faster code generation doesn’t lead to fewer features being built. It leads to more features, requested faster. Better slide production doesn’t lead to fewer presentations. It leads to more presentations, demanded more frequently.
The productivity gains at the individual task level are real. But if the broader organisational workflow becomes choked with excess, low-value output, systemic productivity doesn’t improve, it degrades over time.
This is why AI adoption without workflow governance, without deliberate decisions about what gets done with the time and capacity that AI frees up, so often fails to deliver the expected returns.
Wellbeing and Legal Considerations for HR
This is where the productivity paradox intersects with employment law and HR compliance.
Duty of care: Employers have a duty to take reasonable steps to prevent foreseeable harm to employee health. If AI implementation is contributing to measurable increases in cognitive load, stress, and burnout, organisations face real exposure under health and safety legislation.
Working time: If AI tools are effectively expanding the scope and volume of work without a corresponding increase in capacity, employees may find themselves regularly working hours that approach or exceed legal limits. This requires monitoring, not just assuming that “AI saves time.”
Constructive dismissal risk: Material changes to working conditions, including the nature and volume of work expected, can give rise to constructive dismissal claims if employees are not properly consulted and supported through transitions.
Reasonable adjustments: For employees with existing mental health conditions, an AI-driven increase in cognitive complexity and monitoring demands may trigger obligations around reasonable adjustments.
Practical Steps to Address the Paradox
Audit how AI time savings are actually being used. Don’t assume they’re translating into rest, strategic thinking, or development. Survey your teams. Measure actual workload change, not just tool adoption rates.
Implement workflow pruning alongside AI adoption. For every task AI takes over, identify what doesn’t need to be done anymore. If AI drafts routine reports faster, does that mean you need more routine reports, or fewer, better ones?
Set clear quality standards for AI outputs. Don’t let employees spend hours fixing bad AI output because there are no clear standards for when AI-generated content is “good enough” to use. Build review protocols. Define accountability.
Monitor wellbeing metrics post-implementation. Track absence rates, engagement scores, and burnout indicators before and after AI rollouts. Treat these as implementation KPIs, not afterthoughts.
Protect cognitive recovery time. Avoid the trap of expecting employees to use AI-freed time purely for more output. Rest, reflection, and creative thinking require unstructured time. That time has measurable value.
The AI Productivity Paradox isn’t inevitable. It’s the predictable consequence of deploying AI without addressing the human systems around it. Time saved by machines doesn’t automatically become meaningful rest or strategic thinking for humans, not without deliberate design.
For HR professionals, this is both a challenge and an opportunity. The organisations that manage AI adoption with genuine attention to employee cognitive wellbeing, sustainable workload design, and clear governance frameworks will get more out of their technology investment, and keep their people in better shape to use it.
Next in the series: Article 6, AI-Enhanced Work: How to Actually Multiply Human Capability (Not Just Add Noise).
Intro: AI & the Future of Work Series
About the author
Mick Lavin is an Intercultural Coach, Executive Coach and Mentor, accredited by the European Mentoring and Coaching Council. For the past 30+ years, Mick has worked in the world of technology as a people, project, and strategic account manager in several European countries, with the US, in the Middle East, and in Asia. Mick specialises in people & leadership development and business agility in multicultural business environments, helping organisations move to a more responsive and people-centric mindset.
