Your AI is creating more work, not less, the 40% rework trap
TL;DR
New data from a Workday survey analysed by CIO magazine shows roughly 40% of AI time savings are immediately wiped out by rework, making AI a net productivity drag in many workflows. Your best employees are absorbing the heaviest burden, losing up to 1.5 weeks a year to AI cleanup work that did not exist before you introduced the tools. Mandating adoption without training makes it dramatically worse: employees who feel forced to use AI produce 65% more workslop, according to a Harvard Business Review study. The fix is not better software, it is measuring net value, investing in real training, and listening to the people actually doing the work.
What is the 40% productivity trap?
The headline number comes from a large Workday survey analysed by CIO magazine: approximately 40% of the time saved through AI tools is immediately offset by the extra work needed to fix AI-generated errors.
For every ten hours your team gains by using AI to draft emails, write code, or summarise reports, four hours disappear into cleanup. That is not a rounding error, it is a structural drain on your productivity budget that most businesses are not measuring.
The phenomenon now has a name: workslop. It refers to the proliferation of low-quality, inaccurate, or subtly flawed content produced by generative AI that requires human intervention to fix. It works brilliantly for simple, repeatable tasks. But for complex, nuanced, or high-stakes work, analyst reports, legal summaries, technical documentation, client proposals, the AI's confident output is often more dangerous than a blank page, because it looks authoritative even when it is wrong.
As one executive vice president at a European tech firm put it, the real challenge is getting far more granular about where AI adds value and where it creates rework.
Why is leadership completely blind to this problem?
Here is the uncomfortable truth about most AI dashboards: they measure gross efficiency, not net value.
As a CEO or business owner, you see high adoption rates. You see thousands of words generated and lines of code written in record time. What you cannot see, because your metrics do not capture it, is the endless cycle of revisions, the frantic fact-checking, and the quiet frustration of the people turning AI hallucinations into something usable.
Metrics aimed at the amount of time AI saves can completely lose sight of the actual quality of the results. Speed is up, but if AI-generated mistakes, revisions, and frustration are also climbing, the tool is adding friction instead of removing it.
A survey highlighted by The Guardian quantified this disconnect precisely:
- 92% of high-level executives say AI makes them more productive
- 40% of non-managers say AI saves them absolutely no time at all
The bosses think it is working perfectly. The workers are drowning in workslop.
Who is actually paying the price for AI rework?
Look at your top performers. The Workday study found that the employees most eager to adopt AI, your most engaged, forward-thinking staff, are bearing the brunt of the rework burden.
- 77% of daily AI users audit AI work with the same or greater rigour than they apply to human work
- The bulk of this added labour costs highly engaged employees 1.5 weeks of lost time per year
Because they understand the technology, they know exactly how wrong it can be. They catch the fabricated statistics, the missing legal nuances, the omitted client details that actually matter. They become the safety net for the entire organisation.
"A company's strongest employees often become the safety net, they're the ones catching mistakes, fixing issues, and making sure things don't slip through the cracks. Over time, that can feel less like high-impact work and more like constant cleanup, which is unsustainable long term."
The destructive downstream effect: your best people stop doing high-impact strategic work because they are too busy proofreading a machine. And they are the most likely to leave when they have had enough. You are not just losing productivity, you are creating the conditions for a talent exodus.
See where AI fits in your business. Free.
A 45-minute audit. We map the highest-value automations and what they're worth in time and money. No pitch, no pressure.
Why does mandating AI adoption make the problem dramatically worse?
Many businesses, desperate to realise the promised ROI on their AI investments, have shifted from encouraging AI use to mandating it, tying usage metrics to performance reviews. The evidence on this approach is now overwhelming.
A Harvard Business Review study looking specifically at employee perception of AI strategies found:
Employees who feel mandated to adopt AI show a 65% higher self-reported rate of producing workslop compared to those who are simply encouraged.
When you force people to use a tool they do not trust or do not know how to use effectively, they do not become more productive. They engage in performative compliance, churning out AI-generated content to hit their metrics, knowing full well it is low quality. The same HBR study found that employees who suspect their organisation's ultimate goal is replacement rather than empowerment are the most likely to produce workslop. The intent behind your AI strategy matters as much as the tools you deploy.
Mandated employees also show a meaningfully higher intent to leave. You are not just killing productivity, you are actively driving your workforce out the door.
Is the AI training gap driving this failure?
Yes, and the data makes it embarrassingly clear:
- 66% of leaders cite AI skills training as a top investment priority
- Only 37% of daily AI users report actually having increased access to training
- 54% of AI users who struggle with the technology say their required skills have not been updated, leaving them unsure of where to even start
Companies are rolling out tools faster than they are teaching people how to use them. They are handing out power tools without a manual and acting surprised when the house ends up crooked.
Effective AI training is not showing people how to log in and write a basic prompt. It means teaching teams how to evaluate output, how to spot hallucinations, and how to integrate tools into specific workflows safely. It means setting clear quality standards and giving people the explicit, stated permission to reject AI output when it is not up to scratch, and making clear that saying "the AI got this wrong" is not a career-limiting move.
How do you measure net AI value instead of gross efficiency?
Stop measuring how fast your team generates a first draft. Start measuring the entire task lifecycle:
- How many revision rounds does the output require?
- How much senior staff time goes into auditing it?
- What is the quality of the final deliverable compared to pre-AI workflows?
If a workflow consistently requires heavy rework, or if your high performers spend more time editing than creating, the AI is adding friction, not value. The businesses getting this right are the ones that have stopped applying AI to everything and started getting granular about where it genuinely helps.
Pull AI out of the workflows where it creates net drag. Double down on the workflows where it creates net gain. That sounds obvious, very few businesses are actually doing it.
What to do this week
- Audit one AI-heavy workflow for net value. Pick the process where your team uses AI most. Map the full lifecycle: generation time, revision rounds, senior review time, final quality. Calculate whether it is a net gain or net loss.
- Ask your non-managers directly. Hold a 20-minute session with the people doing the work, not the people managing it. Ask where AI is genuinely helping and where it is slowing them down. Believe what they tell you.
- Review your training provision. If fewer than half your AI users have received updated, role-specific training in the last six months, that is your most urgent fix, before you roll out any new tools.
- Remove any AI usage mandates tied to performance reviews. Replace them with quality outcome metrics. Measure what the AI produces, not how often it is used.
- Identify your safety-net employees. Find the top performers quietly absorbing the rework burden. Acknowledge the cost directly and make a concrete plan to reduce it.
Where to from here
Book a free 60-minute AI audit, we'll explore exactly what workflows are worth augmenting with AI.
Live with passion & AI,
Brett
Want this installed in your business?
Bespoke AI implementation across your operations: strategy, build, rollout, and ongoing drift maintenance.
Frequently asked questions
What is the 40% AI productivity trap?
+
According to a Workday survey analysed by CIO magazine, roughly 40% of the time saved by AI tools is immediately offset by the extra work needed to fix AI-generated errors. For every ten hours of efficiency gained, four hours are lost to cleanup.
What does 'workslop' mean?
+
Workslop refers to the proliferation of low-quality, inaccurate, or subtly flawed content produced by generative AI that requires human intervention to fix. It has become a recognised phenomenon as AI adoption has scaled across businesses.
Do executives know AI is creating more work for their staff?
+
Largely, no. A survey highlighted by The Guardian found 92% of high-level executives say AI makes them more productive, while 40% of non-managers say it saves them no time at all. Leadership dashboards show gross efficiency but are blind to net value.
Why does mandating AI adoption make things worse?
+
A Harvard Business Review study found employees who feel mandated to adopt AI show a 65% higher self-reported rate of producing workslop than those who are simply encouraged. Forced adoption produces performative compliance, churning out low-quality AI output just to hit metrics.
How much time do top performers lose to AI rework each year?
+
The Workday study found highly engaged daily AI users lose the equivalent of 1.5 weeks per year to AI cleanup work, time that did not exist as a cost before AI tools were introduced.
What is the AI training gap?
+
66% of leaders cite AI skills training as a top investment priority, yet only 37% of daily AI users report actually receiving increased access to training. 54% of AI users who struggle with the technology say their required skills have not been updated.
How should businesses measure AI productivity properly?
+
Instead of measuring gross efficiency, how fast AI generates output, businesses should measure net value across the full task lifecycle, including revision rounds, senior staff audit time, and final output quality. If a workflow consistently requires heavy rework, AI is adding friction, not value.
Brett is a four-time founder (Darra Tyres, Gladfish, EzyTrac, Anaboo) and the operator behind AIOS, Anaboo's AI Operating System. He writes from inside the build, installing AI in his own businesses first and reporting back what actually moves the numbers. Based between Singapore, the UK and Australia.
