08.07.2026
Ask most organisations what they offer someone coming back from a year out (parental leave, a caring responsibility, cancer treatment, burnout recovery) and you get a familiar answer.
A phased return. A couple of “keep in touch” days. A refresher session. A buddy for the first fortnight. Warm, well-meant, and built for a world that no longer exists.
That playbook assumes the workplace someone left is broadly the workplace they will return to a recognisable way of working, waiting for them, with a few new tools bolted on. For most of the last thirty years, that assumption held. It doesn’t anymore.
I work at the intersection of AI and human partnering. My research and advisory work is concerned with one question: not what AI can do, but how people experience and perform alongside it. The pattern I keep seeing is this. Career breaks haven’t got harder because people have changed. They feel harder because the workplace is now AI-mediated, and re-onboarding culture hasn’t kept up.
This is a commercial problem as much as a cultural one. If you are trying to retain experienced people, widen your talent pool, or raise female representation, an outdated return-to-work model is quietly working against you. And the people it costs you are the ones you spent years and money developing.
What changed: the half-life, not the person
The usual story about return-to-work difficulty puts the burden on the individual. Confidence dips. Skills go rusty. You need to “get back up to speed.” All true, and all manageable in a workplace that changes gradually.
The World Economic Forum’s Future of Jobs Report 2025, drawing on more than 1,000 employers across 55 economies, puts a number on how much that assumption has shifted. Employers expect 39% of workers’ core skills to change or become outdated by 2030. That figure has eased from earlier editions, down from 44% in 2023 and a pandemic-era peak of 57% in 2020. Even so, it means roughly two of every five skills a person holds today will be reshaped inside five years. Analysts tracking technical roles put the effect more sharply: the half-life of many technical skills is now below two and a half years.
Sit with that in the context of a career break. If a meaningful share of skills turns over inside two to three years, then a twelve- or eighteen-month absence is no longer a gap in a stable system. It is a gap during structural reconfiguration. The person didn’t fall behind. The ground moved.
AI is the thing moving it fastest. AI and big data top the WEF’s list of fastest-growing skills, and skills gaps are now the single biggest barrier to business transformation, cited by 63% of employers, ahead of culture, capital, and regulation. This is why “you’ll slot right back in” is breaking as a promise. A year out can now mean returning to a different operating model, not a familiar role with a few new dashboards.
The second layer: algorithmic anxiety on top of technostress
Here is where the human experience diverges from the org chart. When someone returns now, they are not only relearning tasks. They are re-entering systems that make decisions.
Technostress, the psychological strain of adapting to workplace technology, is well established in the research literature and predates the current AI wave. What has been added on top is narrower and more corrosive: algorithmic anxiety. A 2026 study in Frontiers in Psychology, analysing over 1,400 first-hand accounts of AI-driven change at work, found that alongside surface optimism there was significant negative sentiment reflecting a deeper concern people described as “algorithmic anxiety” related to job loss. The same body of work describes how AI-driven decisions feel impersonal and arbitrary, lacking the human element that traditionally cushioned difficult organisational transitions, and how algorithmic management can leave workers feeling reduced to data points, with heightened cynicism and technostress.
Researchers have given the underlying mechanism a name: “algorithmic accountability gaps,” where workers cannot identify who to hold responsible for AI-driven decisions affecting their livelihood. When the rules are encoded in a model rather than held by a person, there is no one to ask. For a returner, that is disorientating in a specific way. The visible things (jargon, dashboards, team members) have changed. The invisible things (how work is routed, prioritised, and evaluated) have changed too, and they are harder to read.
This is why I describe the returner’s experience as psychological whiplash, not simple re-onboarding nerves. It is not one adjustment. It is layered: relearning the role, decoding new systems, and doing both while sensing that the terms of the job itself have shifted. This apprehension is not confined to the person returning. In the WEF data, resilience, flexibility and agility, along with leadership, rank among the most-valued core skills precisely because employers are themselves uncertain about the skills the future will demand. The manager welcoming someone back is often navigating the same anxiety. The returner walks into a team where even the leadership is unsure.
Why the old playbook fails
Phased returns, keep-in-touch days, and short refreshers were designed for linear change. They assume the difference between “before” and “after” is small enough to close with time and reassurance. They were never designed to absorb structural reconfiguration of the underlying work.
There is nothing wrong with those mechanisms. The problem is that they are treated as sufficient, and as administrative rather than developmental. A keep-in-touch day that genuinely kept you in touch made sense when a year’s change amounted to a new intranet and a reshuffle. On its own, it does not bridge a return into algorithmically mediated systems and AI-reconfigured decision-making.
If AI compresses the pace of change, then the human infrastructure around it (culture, onboarding, psychological safety) must expand to absorb the compression. Re-onboarding must become a deliberate capability intervention rather than a form to complete. In practice, that means four shifts.
Re-immersion, not admin. Treat the return as a genuine re-entry into a changed operating environment, with structured exposure to how work now flows, rather than a desk and a login.
Psychological safety first. The research is consistent that anxiety about AI drives avoidance and disengagement. One study found that higher AI anxiety increases people’s intention to stop using the technology altogether. A returner who feels judged for not knowing the new tools will hide the gap rather than close it. Safety is the precondition for learning, not a nicety added on top.
AI literacy as baseline, not bonus. If AI now mediates the work, then understanding what these systems do, and where they are unreliable, is core to the role. It cannot be an optional module the confident opt into.
Manager readiness, not just employee resilience. We routinely ask the returning individual to be resilient. We rarely equip the manager to lead a return into a system they are themselves still learning. That asymmetry is where good intentions quietly fail.
The hidden talent drain, and who it hits
The uncomfortable arithmetic is this. Time away has become more expensive in capability terms than it was even three to five years ago, because the rate of change underneath it has risen. If re-onboarding doesn’t evolve, organisations don’t lose people loudly. They lose them quietly, through people who return, feel they can no longer read the place, and decide it can no longer read them.
That loss is not evenly distributed. Career breaks fall disproportionately on caregivers, parents, and those navigating serious health events, and in the UK that population skews heavily female. Government-cited research puts around 427,000 UK women with professional backgrounds currently on career breaks and looking to return. The Institute for Fiscal Studies has shown the penalty is already real and cumulative: hourly wages for women who return are on average about 2% lower for each year taken out, rising to 4% per year for women with at least A-level qualifications. The screening bias compounds it. Harvard Business School research found that 43–48% of employers with applicant tracking systems filtered out skilled candidates with CV gaps of over six months on that basis alone.
Now layer AI exposure on top of an already-penalised group. The evidence here is unusually clear. The ILO’s 2025 analysis with Poland’s NASK institute found that women are disproportionately more likely than men to be in roles highly exposed to generative AI, and that if the most-exposed jobs disappeared, two women would be displaced for every man. The 2026 follow-up sharpened it: around 29% of female-dominated occupations are exposed to GenAI, against 16% of male-dominated ones; at the highest-risk end, 16% of female-dominated roles against just 3% of male-dominated ones.
So, the people most likely to take an extended break are also the people most likely to return into roles being reshaped by AI, and least likely, per the same research, to have had AI training along the way. For any organisation working to raise female representation, that is not a peripheral HR concern. It is a direct threat to the goal, operating through a process most leaders don’t currently look at: how people come back.
The through-line: this is a human-premium problem
It would be easy to read all of this as an argument for more technology training. It isn’t. The evidence points the other way.
The WEF’s own data is explicit that as technical skills churn, employers will demand human-centric skills (judgment, problem-solving, collaboration) to work effectively with the technology, and that the longevity of technical skills is shortening while workforce readiness lags. The scarce, appreciating asset in an AI-mediated organisation is not the tool. It is the experienced human who can judge when to trust the tool and when not to, the person whose value is that they hold context, pattern recognition, and accountability that a model does not.
That is the person a career break puts at risk of walking out the door. Which reframes re-onboarding entirely. It is not a welfare gesture or a diversity checkbox. It is how you protect the highest-value, hardest-to-replace capability you have, now AI makes that capability more valuable rather than less. The organisations that hold onto experienced people through the AI transition will be the ones that treat the human return as seriously as the technology rollout.
The alternative is a slow, invisible loss of the people you can least afford to lose, each one leaving with some version of the same quiet verdict:
“I don’t recognise this place anymore. And I’m not sure it recognises me either.”
If AI is going to compress the pace of change, the answer is not to ask people to absorb the shock alone. It is to build a return that expands to meet them, because the capability walking back through the door is worth more now than it has ever been.
Where this becomes practical
If any of this maps onto what you are seeing (returners who don’t stay, a re-onboarding process that stops at admin, a female-representation target that keeps stalling for reasons nobody can quite locate) it is measurable, and it is fixable.
This is the work I do at Morson. My team runs evidence-led diagnostics on AI readiness and techno social response, then translates the findings into re-onboarding models, workforce strategies, and leadership interventions built for the pace of change your organisation is facing, not the one the old playbook assumed. It combines behavioural science with applied advisory across regulated, high-stakes sectors including infrastructure, energy, aerospace, and defence.
