Experience debt: what happens to judgement when AI takes the tasks people learned from
· 5 min read
New research finds most people say skills are slipping as AI takes on routine tasks, while time to learn is scarce. What that means for leaders, and which toolkit tools help rebuild the learning route.
In brief
AI is taking over many of the routine tasks that people once learned their jobs from, and new research suggests organisations are not yet replacing that learning. In a Dayforce survey of almost seven thousand workers, managers and executives in six countries, including the UK and Germany, nearly two thirds said people are losing the ability to do some tasks themselves as AI takes them on. Around half said the tasks that used to build experience are now handled by AI. Workers also report too little time to learn new skills on top of their normal workload, while executives are far more likely to think that time exists. A separate UK study from O'Reilly found many workers now ask a chatbot first and a sizeable share have acted on AI answers that proved wrong. Both studies come from vendors and report what people say rather than measured skill loss. This article explains what the findings mean for leaders, why routine work has always been where judgement is built, and how tools from the AI Enablement Toolkit for Leaders can help a team decide what each task was teaching and how that learning will happen now. It also covers where training for managers fits and the limits of what the toolkit can do.
Ask anyone good at their job how they got good, and the answer is rarely a course. It is usually a run of small, unglamorous tasks: the reconciliation that would not balance, the draft that came back with red ink, the customer call that went wrong and had to be put right. That work produced output. It also produced judgement.
New research suggests that second product is quietly disappearing as AI takes over the first.
What the research found
Dayforce, which sells workforce and HR software, released its 17th annual Pulse of Talent report, "Fast now, fragile later", on 26 October 2026. This is vendor research. Hanover Research surveyed 6,909 workers, managers and executives online between 27 July and 14 August 2026, at organisations with more than 100 staff in Australia, Canada, Germany, New Zealand, the United Kingdom and the United States.
The findings that matter most for leaders:
- 65% say people are losing the ability to perform some tasks themselves as AI takes them on.
- 49% say tasks that traditionally helped employees build work experience are increasingly handled by AI, and 51% say AI is reducing early-career opportunities.
- 68% say organisations are losing important skills or expertise without fully realising it, and 73% say lost capabilities are harder to rebuild than organisations expect.
- 50% of workers say they are expected to develop new skills while keeping the same workload, but only 27% say they have enough time at work to do so. 56% of executives say employees do have enough time.
- Workers want AI-saved time used to reduce workload and burnout (64%) and for learning and development (50%), but 40% expect it to become more work.
Dayforce calls the result "experience debt", which it describes as a future capability cost that accumulates when people lose opportunities to learn by doing.
A separate UK study points the same way from the other end. O'Reilly, which sells training, released research on 27 October 2026 (Censuswide, 2,000 UK employees). It found 39% of UK workers now turn to an AI chatbot first for work answers, 35% have had no AI training or guidance in the past year, and 39% of those using AI answers have acted on one that later proved inaccurate, incomplete or unsuitable. Only 24% always verify.
Put the two together and a pattern appears. The tasks that taught people to spot a wrong answer are going to the machine, while the machine's answers still need someone who can spot a wrong one.
What it means for leaders
None of this is a case for keeping work people would rather not do. Much of the grind deserves to go. The problem is that the grind was doing two jobs, and most business cases only count one.
Ian Hughes makes the point in What Stays the Same, writing about his own start as a junior programmer at Marks and Spencer:
"The junior work was never just output. It was the school. It was where judgement was quietly being built, one unglamorous task at a time."
He is also clear about the consequence: "The thing to protect is the apprenticeship, not the task."
The gap between executives and workers on time is the part leaders can act on fastest. If 56% of executives believe people have time to learn and only 27% of workers agree, the learning is not happening on the schedule leaders think it is. And if AI-saved time flows straight back into more work, nothing has been set aside to rebuild what the old tasks used to teach.
Which toolkit tools help
The AI Enablement Toolkit for Leaders is a leadership method rather than software. Several of its tools fit this problem directly.
- Tool 14, Protect the apprenticeship: for each task AI may remove, name the capability you will need in future, the old task that helped build it, the new safe learning experience that replaces it, and who coaches it. Its design principles include using AI outputs "as material to critique, not as an answer key".
- Tool 7, See the work: follow real cases, not the documented happy path, so you can see where people actually learn, check and correct.
- Tool 10, Human and AI role design: make explicit where a person must review before acting, and what they need to know to do that well.
- Tool 18, Proof-of-value scorecard: treat saved time as capacity released, which needs a plan for where it goes, not as cash saved.
- Tool 13, Incentive and behaviour audit: if learning is praised but only output is rewarded, people will choose output.
- Tool 20, Learning review: build a regular point where the team asks what it has learned, not only what it has shipped.
Where training fits
Rebuilding the learning route is a leadership capability. Managers who learned by doing now need to coach judgement on purpose: setting work where newer colleagues can be wrong safely, reviewing AI output with them rather than for them, and explaining what good looks like. Esm training and enablement can help leaders and managers build that skill, alongside role-specific AI training for the people doing the checking.
Honest limits
Both studies are vendor research from companies with an interest in the answer, and both report what people say rather than measured skill loss. The Dayforce figures mix workers, managers and executives across six countries, and its public release does not break the results down by country. The toolkit cannot measure skill decline for you, and it is not a learning curriculum. It gives a leadership team a structured way to decide, task by task, what the old work was teaching and how that learning will happen now.
The question to take into your next AI business case is a simple one: when this task goes to the machine, where will the judgement it used to build come from?
Tags: skills, experience-debt, apprenticeship, leadership, ai-adoption
Sources
Join the Exchange free
Join free to unlock the complete knowledge library.
Founder membership is free for life
More from Esm
How do I find everyday AI uses for staff outside IT?
5 min read
How should IT teams handle emergency patches for critical vulnerabilities?
5 min read
When the human in the loop becomes a rubber stamp: what UK regulators found about AI checks
5 min read
10 scheduled AI tasks for service management teams
8 min read
Boards expect AI savings: why almost no organisation is finding them
5 min read
AI rebounding: why some employers are asking back the people AI replaced
5 min read
Related articles
How do I find everyday AI uses for staff outside IT?
Start with the work your teams already do, not a list of AI tricks: observe, set data rules, run small timed trials with an owner, and publish what works as knowledge.
How should IT teams handle emergency patches for critical vulnerabilities?
Treat an emergency patch as an emergency change: triage the risk fast, agree the fast path in advance, test as far as the clock allows and keep a remediation plan.
When the human in the loop becomes a rubber stamp: what UK regulators found about AI checks
The Bank of England and FCA find most firms approve AI before launch but far fewer keep watching it, and the FCA warns human review can become a tick-box exercise. What leaders should check, and which toolkit tools help.
Boards expect AI savings: why almost no organisation is finding them
Gartner finds 77% of board members expect AI to deliver savings, but only 3% of organisations beat their savings goals through productivity. What leaders should agree instead, and which toolkit tools help.
AI rebounding: why some employers are asking back the people AI replaced
New SHL research finds some UK and US workers replaced by AI were later asked back. Why it happens, and how leaders can understand the work before automating it.
AI Job Cuts and the Unfilled Role: Where Should Freed Capacity Go?
Barclays' AI-assisted job cuts and a quieter freeze on junior hiring raise one leadership question: when AI frees capacity, who decides where it goes before the spreadsheet decides for you?