#THINKFUTURE
Series: Building the Future on a Cracked Foundation
About this series
Artificial intelligence has become one of the defining leadership topics of our time. Organisations are investing billions in technologies that promise greater productivity, lower costs and faster decision-making. Those investments matter, but this series begins with a different question:
What if the greatest challenge of AI is not how it changes work, but how it changes the way organisations develop capable people?
In this THINK FUTURE series, Building the Future on a Cracked Foundation, we explore that question through the lenses of research, organisational psychology and practical leadership. Each essay challenges a common assumption and invites leaders to think differently about capability, work and organisational success.
The graduate intake that no longer made financial sense
Imagine the executive committee of a professional-services organisation reviewing its workforce plan.
For years, the organisation has recruited a sizeable group of graduates. They entered as analysts, trainees and junior consultants. Their first assignments involved gathering information, checking data, preparing initial drafts, documenting processes and supporting more experienced colleagues.
The work was not always exciting. Some of it was repetitive. Much of it required correction. Junior employees consumed managerial time before they produced work that could be trusted independently.
Generative AI has changed the calculation.
The organisation can now complete much of the research, analysis and drafting with fewer people. Experienced professionals use AI to prepare first versions of reports, review large volumes of information and explore alternatives in a fraction of the time previously required.
The financial case appears compelling. Why recruit twenty graduates when twelve AI-enabled professionals could produce more work, sooner and at a lower cost?
The workforce plan is revised. Graduate recruitment is reduced. Several junior positions are left unfilled. The immediate effect is positive: productivity rises, margins improve and managers spend less time correcting elementary work.
Five years later, the organisation needs experienced specialists who understand its clients, its methods and the contexts in which its advice must be applied. It also needs managers capable of supervising complex engagements and developing the next generation of professionals.
The external market contains experienced candidates, but they are expensive, scarce and unfamiliar with the organisation. Internally, too few people are ready to progress. The organisation has improved the efficiency of today’s work while narrowing the pathway that once produced tomorrow’s expertise.
The scenario is hypothetical. The tension is not.
It raises a question that reaches beyond the future of graduate employment:
What happens to an organisation’s future specialists, managers and leaders when it no longer needs as many people at the beginning of their careers?
The assumption built into the career ladder
The familiar career ladder rests on an assumption that is rarely made explicit: careers will continue developing through broadly the same sequence of roles and experiences through which previous generations progressed.
People enter at the bottom, learn the fundamentals, demonstrate competence and assume progressively greater responsibility. Some become specialists. Others become managers or leaders. The organisation recruits repeatedly at the base because it expects some of those employees eventually to occupy positions higher in the structure.
The ladder has never worked perfectly. Careers are not universally linear, promotion has never been equally accessible and time served does not necessarily produce expertise. Some employees remain trapped in junior work that offers repetition without meaningful development. Others leave because the next rung is unavailable or because the organisation fails to recognise their potential.
Nevertheless, graduate programmes and junior roles have performed a function that is larger than the completion of lower-level tasks.
They have brought inexperienced people into organisations and exposed them to real work. They have allowed new professionals to observe how experienced colleagues think, learn what quality looks like, receive correction and gradually become trusted with more difficult decisions.
The career ladder was therefore never only a hierarchy of positions.
It was also a developmental system for converting inexperienced entrants into experienced professionals.
AI challenges that system because many of the tasks most amenable to automation are concentrated near its base.
The evidence is emerging, not settled
Concern about the disappearance of entry-level work has grown rapidly, but the evidence requires careful interpretation.
A 2025 Stanford working paper used payroll data covering millions of workers in the United States to examine employment changes following the widespread adoption of generative AI. The researchers reported a 13 per cent relative decline in employment among workers aged 22 to 25 in the most AI-exposed occupations, after controlling for firm-level shocks. Employment for more experienced workers in the same occupations remained stable or continued to grow. The declines were concentrated in occupations where AI appeared more likely to automate work than augment it. The authors describe their findings as early evidence consistent with the hypothesis that generative AI is disproportionately affecting young workers. The study remains a working paper, is based on one national labour market, and does not establish that AI alone caused every observed change.
Other evidence complicates a simple displacement narrative. A 2026 survey of senior talent leaders in the United States found that employers using or exploring AI were more likely to report an increase than a decrease in entry-level hiring. Nearly three times as many expected AI to increase entry-level hiring in 2026 as expected it to reduce it. Respondents also reported that AI was increasing the analytical and judgement-related content of some entry-level jobs.
These findings are not necessarily contradictory. AI may reduce entry-level employment in some occupations and organisations while increasing it in others. It may replace particular tasks, make junior workers more productive, create new forms of work, or change the level at which organisations consider a person ready to contribute.
Broader research from the International Labour Organisation also cautions against treating task exposure as automatic job elimination. Its 2025 global index concluded that job transformation is the more likely general effect of generative AI because most occupations contain tasks requiring continued human involvement. Exposure is nevertheless uneven, with clerical work among the most affected categories.
The evidence therefore does not justify declaring that the career ladder has already ended.
It does justify asking whether its first rungs are becoming narrower, steeper or fundamentally different.
A particularly serious question for South Africa
This question has particular significance in South Africa, where entry into work is already difficult for millions of young people.
According to Statistics South Africa, 4.7 million people aged 15 to 34 were unemployed in the first quarter of 2026. The unemployment rate was 60.9 percent among those aged 15 to 24 and 40.6 percent among those aged 25 to 34. More than four in ten young people aged 15 to 34 were not in employment, education or training. Young workers were also underrepresented in managerial, professional and technical occupations.
AI did not create this crisis. South African youth unemployment predates the widespread availability of generative AI and reflects structural economic, educational and labour-market conditions that cannot be reduced to technology.
AI may nevertheless alter the pathways through which young people could otherwise have entered organisations.
In a labour market where employers already prefer experience and face limited capacity to train new entrants, technology that enables experienced employees to produce more without junior assistance may strengthen the temptation to recruit fewer inexperienced people. The individual organisation may regard this as a rational response to cost and productivity pressures.
When many organisations make the same decision, however, the wider consequence may be fewer opportunities for people to obtain the experience employers subsequently demand.
South Africa already confronts this circular problem. Young people struggle to find employment because they lack experience, yet they cannot acquire experience without an opportunity to work. Statistics South Africa reported that 58.7 percent of unemployed young people in the first quarter of 2025 had never worked before.
If AI increases the value of experience while reducing some of the opportunities through which experience begins, the entry barrier may become even more difficult to cross.
Junior work did more than produce junior outputs
The work assigned to inexperienced employees has often been dismissed as low-value work. From a narrow productivity perspective, the description may be correct.
A junior analyst gathers data more slowly than an experienced colleague. A graduate consultant spends hours preparing a first draft that requires extensive revision. A trainee accountant performs reconciliations that increasingly can be automated. A junior lawyer reviews documents that software can process more quickly.
Yet the organisational value of a task and its developmental value are not always the same.
The junior analyst who gathers data begins to understand where information comes from, which inconsistencies recur and why apparently comparable measures may mean different things. The graduate consultant preparing a first draft learns how evidence must be structured into an argument. The trainee accountant performing reconciliations begins recognising what normal transactions look like and which discrepancies require investigation. The junior lawyer reviewing documents encounters the language, patterns and exceptions that later support professional judgement.
Not every repetition produces learning. Poorly designed junior work may teach little more than endurance. Repetition becomes developmental when employees understand what they are doing, receive credible feedback and encounter enough variation to recognise both patterns and exceptions.
This is why work and learning cannot be treated as entirely separate systems.
A 2025 systematic review of 73 studies described workplace learning as the intertwining of working and learning. It identified task complexity, task variety, autonomy, successes and failures, managerial support and organisational learning culture among the conditions shaping whether learning occurs through work.
Junior work provided more than output because it placed employees inside a context where learning could occur. They encountered real clients, organisational constraints, imperfect information and consequences that no classroom could reproduce fully.
If the task disappears, the organisation does not automatically lose the learning. It does, however, need to know where that learning will now come from.
Experience is accumulated through encounters, not years
Organisations often use years of experience as a convenient indicator of capability. A vacancy may require three, five or ten years in a particular field, as if professional growth could be measured by time alone.
Time is an imperfect proxy.
Two people can spend the same number of years in similar positions and develop very different levels of expertise. One may repeat familiar tasks without receiving useful feedback. The other may encounter varied situations, observe capable colleagues, reflect on mistakes and assume progressively greater responsibility.
What matters is not simply how long someone has worked. It is what the person has encountered, attempted, understood and learned to judge.
Early-career roles have traditionally enabled this accumulation through repeated exposure. Junior employees see many examples of routine work. This helps them establish what normal looks like. They then begin encountering anomalies and exceptions. With guidance, they learn which differences matter and what consequences may follow.
Over time, individual assignments accumulate into patterns. Feedback becomes connected to context. Rules become qualified by experience. The person becomes less dependent on explicit instructions because previous encounters have changed how the next situation is interpreted.
AI can accelerate parts of this process. It can expose employees to more examples, simulate cases and provide explanations. As Essay 4 argued, it can strengthen expertise development when it enriches practice, feedback and reflection.
Yet simulated exposure and completed outputs are not automatically equivalent to accountable participation in real work. The employee must still learn what it means to make a recommendation to a client, defend an assumption, respond when a decision fails, and recognise when apparently minor details carry significant consequences.
Experience is not merely information about what happened.
It is participation in situations where judgment matters.
When the beginner is expected to arrive experienced
As AI absorbs parts of junior work, organisations may respond by redefining what entry-level means.
A graduate who once began by collecting information may now be expected to interpret AI-generated analysis. A junior consultant may be expected to validate evidence and engage with clients earlier. A new software developer may spend less time writing elementary code and more time reviewing, integrating, and testing AI-generated code. A trainee may move more quickly from task execution to supervising technological output.
This can be a valuable change. Removing low-value activity may allow early-career employees to engage with more meaningful work sooner.
It also creates a developmental paradox.
Interpretation, validation, and supervision are often treated as higher-order responsibilities because they require sufficient domain knowledge to recognise errors and exceptions. If AI performs the foundational task before the employee understands it, the beginner may be asked to exercise judgment that the task previously helped develop.
The organisation may therefore expect entry-level employees to arrive with experience that can normally be acquired only after entry.
A qualitative study published in 2025 examined how entry-level consultants were reshaping their work through generative AI. The researchers found that these professionals altered not only their tasks, but also their relationships and the ways in which they demonstrated value. The study illustrates why AI’s effect on early careers cannot be reduced to job removal. Entry-level professionals may actively redesign their roles, engage with more advanced work and seek new ways to establish credibility.
The opportunity is substantial, but it depends on support. More advanced assignments given earlier do not automatically accelerate development. Without foundational knowledge, supervision and feedback, they may merely expose beginners to greater complexity before they can interpret it.
A shorter first rung is useful only if the person can still reach it.
Graduate programmes are capability pipelines
Graduate programmes are often evaluated through recruitment numbers, completion rates, retention and the speed at which participants become productive.
Those measures matter, but they may obscure the programme’s larger purpose.
A well-designed graduate programme creates a managed transition between education and professional contribution. It allows the organisation to select for potential, develop context-specific capability and observe employees across varied assignments before making longer-term placement decisions.
Rotations expose graduates to different parts of the system. Cohort learning creates peer support. Structured supervision enables mistakes to be corrected before responsibility becomes excessive. Access to experienced colleagues helps graduates understand standards that may never be fully documented. Progressive assignments reveal who is ready for greater complexity and who requires further development.
Graduate programmes also create an internal evidence base. Organisations learn what potential looks like in their own context, which experiences accelerate readiness and where development frequently stalls. This contributes to better talent decisions, succession planning and workforce capability.
If AI is introduced only as a productivity technology, these programmes may begin to look unnecessarily expensive. The organisation sees the cost of training, managerial attention and lower initial productivity, while AI makes it possible to meet immediate output requirements with fewer people.
If the programme is viewed as a capability pipeline, the calculation changes.
The relevant question is no longer only how much productive work graduates deliver in their first year. It is whether the organisation is building the specialists, managers and leaders it will require in five or ten years.
This does not mean graduate intake should remain unchanged regardless of future demand. Organisations should not recruit people into obsolete roles merely to preserve a familiar structure. It means that reductions in early-career recruitment should be evaluated against future capability requirements, not only current labour efficiency.
Fewer juniors do not mean fewer future experts are needed
The logic of reducing junior employment may seem straightforward.
If AI allows one experienced professional to complete work that previously required several juniors, the organisation needs fewer people at the bottom. If this pattern continues, perhaps it will also need fewer people in the middle and at the top.
That outcome is possible in some areas. It is not inevitable.
AI may reduce the number of people required to produce a particular service while increasing the number of decisions, clients or complex problems the organisation can address. Experienced employees may supervise more AI-enabled activity, but they must still evaluate exceptions, manage relationships, integrate different forms of evidence and accept accountability for consequential decisions.
Organisations may therefore become thinner at the base while remaining dependent on experienced capability in the middle.
This creates what might be called a pipeline compression risk. The organisation recruits fewer entrants but continues expecting a sufficient number of experienced specialists to emerge later.
“Pipeline compression” is an editorial description of the risk, not an approved ECCSA model or established measure. It should therefore be treated as a question for organisational investigation.
The arithmetic is important. Not every graduate remains with the organisation. Not every junior becomes a specialist. Not every high performer wishes to become a manager. People leave, change careers, relocate or discover that a different form of work suits them better.
A broad entry cohort historically gave organisations options. If that cohort narrows substantially, the margin for attrition also narrows. A capability shortage may become visible only years after the recruitment decision that helped create it.
External hiring cannot resolve every gap. Organisations compete for the same experienced people, and external recruits still require time to acquire organisational context, relationships and institutional knowledge. When every organisation seeks experienced talent while fewer are willing to develop it, the market is asked to supply expertise that the market itself has not produced.
The traditional ladder should not be romanticised
The answer is not to preserve every feature of the old career ladder.
Traditional pathways often required junior employees to spend years proving themselves through long hours, repetitive assignments, and limited autonomy. Access depended on educational privilege, social networks, geography and a manager’s willingness to advocate for an employee. Promotion sometimes rewarded conformity more reliably than capability.
Some junior tasks existed because experienced people preferred not to perform them, not because they had been designed deliberately for learning. Employees were expected to acquire professional judgement through exposure, but the quality of supervision and feedback varied considerably.
The ladder also implied that progress meant moving upward into management. This undervalued specialist careers and created pressure to promote technically capable people into roles for which they were neither motivated nor prepared.
AI creates an opportunity to redesign these weaknesses.
Routine administration can be reduced. Early-career employees can gain access to richer problems sooner. Simulations can provide exposure to situations that occur too infrequently in ordinary work. AI can offer explanations and immediate feedback between conversations with managers. Digital work can make employees’ reasoning more visible and help identify where development is needed.
Career pathways can also become more varied. Employees may progress through projects, professional milestones, specialist pathways, rotations and demonstrated capability rather than waiting for a vacancy on the next hierarchical level.
The end of the traditional ladder would not necessarily mean the end of career development.
The danger would be ending the ladder without building a better way to climb.
From a career ladder to a capability pathway
A ladder suggests a fixed sequence. Each rung is occupied before the person reaches the next. The structure is visible, vertical and largely the same for everyone.
Future careers may look less like this.
People may enter through different routes, develop through varied combinations of experience, and move between specialist, project, leadership, and advisory responsibilities. AI may enable employees to attempt more demanding work earlier, but readiness will still need to be demonstrated rather than assumed.
A capability pathway would focus less on time spent in a role and more on the development of the knowledge, judgment, and responsibility required for future contribution.
This does not redefine the approved ECCSA Future Capability Architecture. It is an editorial way of interpreting how career development may need to change in response to AI-enabled work.
Such a pathway would ask:
- Which experiences develop the capability required at the next level?
- Which foundational tasks can be automated without weakening understanding?
- Which tasks must employees still perform, observe or explain before they can judge AI-generated work?
- How will employees encounter sufficient variation, exceptions and consequences?
- Who will provide feedback and determine readiness?
- How will specialist development be supported alongside leadership development?
- What evidence will demonstrate that AI-enabled performance has become durable human capability?
The emphasis shifts from preserving roles to preserving developmental outcomes.
An organisation may no longer need a junior employee to spend hundreds of hours compiling standard reports. It may still need that employee to understand how information was selected, what assumptions shaped the report, which anomalies require investigation and how the recommendation could affect a client.
The old task may disappear. The learning requirement does not.
Redesigning the first rung
Leaders do not need to choose between preserving obsolete junior work and abandoning early-career development.
They can redesign the first stage of professional work around the realities of AI-enabled performance.
Identify the hidden curriculum of junior work
Before automating or removing a task, organisations should determine what employees learned while performing it.
The learning may include understanding data sources, recognising quality standards, observing stakeholder behaviour, practising professional communication or seeing how decisions move through the organisation.
If the learning matters, it must be preserved through another experience.
Recruit for potential, not premature mastery
As entry-level jobs become more complex, organisations may respond by increasing experience requirements. This can exclude the very people the role is meant to develop.
Recruitment should distinguish between capabilities that must be present at entry and those the organisation has a responsibility to develop. Curiosity, learning agility, ethical judgement, foundational knowledge and the ability to work with evidence may matter more than prior mastery of every task.
Give graduates real work with bounded responsibility
Simulations and training exercises are valuable, but they should not become permanent substitutes for meaningful contribution. Early-career employees need exposure to real organisational problems, decisions and relationships.
The responsibility should be consequential enough to create learning but sufficiently supervised to protect clients, colleagues and the organisation.
Use AI to deepen the work
AI can generate alternative scenarios, expose employees to exceptions, challenge initial conclusions and provide immediate explanations. It can allow graduates to compare their thinking with multiple possible approaches rather than merely repeating one established method.
Its role should be to widen and deepen the developmental experience, not only complete the assignment faster.
Make observation reciprocal
Junior employees need opportunities to observe experienced professionals framing problems, managing uncertainty and responding when information is incomplete.
Senior employees also need to observe how juniors reason with AI. A polished final output is insufficient. Development becomes more visible when employees explain their assumptions, identify what the AI contributed and defend decisions to knowledgeable colleagues.
Build rotations around learning, not movement
Rotations become valuable when each assignment develops a distinct form of knowledge or judgment. Moving employees between departments without clear developmental outcomes may create breadth without depth.
Each rotation should contribute deliberately to the capabilities required for a future specialist, manager or leader.
Create credible specialist pathways
Not every capable employee should have to enter management to progress. Organisations will continue to need people with deep technical, professional and contextual expertise.
Specialist pathways should offer progressively greater complexity, influence, recognition and reward. Otherwise, organisations may lose precisely the experts required to oversee AI-enabled systems and teach developing professionals.
Measure readiness rather than attendance
Completing a graduate programme or spending a specified number of years in a role should not automatically demonstrate readiness.
Employees should be able to explain reasoning, transfer learning, recognise exceptions, manage uncertainty and accept progressively greater accountability. These provide stronger evidence of development than tenure alone.
The decisions leaders are really making
When executives decide to reduce graduate recruitment or automate junior work, they may believe they are making a resourcing decision.
They are also making a decision about the organisation’s future supply of capability.
The consequences will not appear immediately. The work may continue. Productivity may improve. Experienced employees may compensate for the smaller intake, particularly when AI increases their capacity.
The effect becomes visible when the organisation needs someone who has accumulated several years of relevant experience, understands the context and is ready to assume responsibility.
At that point, leaders may discover that they did not eliminate the need for the middle of the career structure.
They reduced the number of people who could reach it.
This does not mean every organisation should maintain the same graduate intake or preserve every junior role. It means that workforce planning must connect present automation decisions with future specialist, managerial and leadership requirements.
A credible AI business case should therefore ask more than:
- How much work can be automated?
- How many people will be required?
- How quickly will the investment pay for itself?
It should also ask:
- Which developmental experiences will disappear?
- Which future roles depend on those experiences?
- How many people must enter now for enough capable people to emerge later?
- How will AI-enabled work develop rather than merely use human capability?
- What alternatives will replace the learning pathways being removed?
These are not arguments against productivity. They are conditions for ensuring that productivity does not consume the capability on which future performance will depend.
A different leadership conversation
Have we reached the end of the career ladder?
Not yet, and perhaps not in the way the question first suggests.
Hierarchical career progression is unlikely to disappear entirely. Organisations will still require different levels of responsibility, expertise and accountability. People will still enter with limited experience and develop towards more demanding contribution.
What may be ending is the assumption that this development will occur automatically through the familiar sequence of graduate recruitment, junior work, time served and eventual promotion.
AI is changing the tasks at the bottom. It may reduce some entry-level opportunities while transforming or expanding others. It may remove repetitive work that added little value, but it may also remove encounters through which beginners learned what normal looked like before they were expected to recognise what was wrong.
The challenge is not to defend every rung of the old ladder.
It is to understand what the ladder was doing before removing it.
Graduate programmes, junior roles and supervised work were not merely labour arrangements. At their best, they were mechanisms through which academic knowledge became contextual understanding, repeated exposure became pattern recognition and guided responsibility became professional judgement.
Future career pathways can improve on that system. They can be more inclusive, more flexible and more deliberately developmental. AI can support richer practice, faster feedback and earlier engagement with complex work.
None of this happens automatically.
If organisations automate the task but neglect the learning, they may improve present performance while weakening the supply of future experts. If they redesign the pathway as deliberately as they redesign the work, AI may shorten the distance between entry and meaningful contribution without removing the experiences required for sound judgement.
THINK FUTURE Principle
The career ladder was never only a route to promotion. It was a system for turning potential into experience and experience into responsible capability.
AI may change the roles, tasks and sequence through which careers develop. The leadership responsibility is to ensure that removing yesterday’s work does not remove the development of tomorrow’s specialists, managers and leaders.
The question is therefore not whether the traditional ladder should be preserved.
It is:
If the first rung disappears, how do people climb?
Coming next
Who Is Teaching Tomorrow’s Experts?
Career pathways do not develop people by themselves.
Historically, experience taught. Managers coached. Mentors shared judgement that could not be found in manuals, and teams helped newcomers understand how professional work was really performed.
AI is now becoming part of this learning environment. It can explain, demonstrate, question, simulate and provide feedback at a scale no human manager can match. It can also give inexperienced employees plausible answers without the contextual understanding required to evaluate them.
In the next essay, we examine how responsibility for developing expertise changes when AI becomes part of the teacher:
Who will teach tomorrow’s experts, and what must remain distinctly human in that relationship?