The Right to Retraining Is Not Enough

By Dr Riaan Steenberg

A company introduces an intelligent system. Parts of a job disappear. The affected employees are offered an online course and told that the organisation is committed to reskilling.

The course is available after hours. Completion is voluntary. No protected learning time is provided. The new roles require experience the employees cannot acquire in their current jobs. Hiring managers still prefer external candidates who have already done the work.

Technically, retraining was offered.

Practically, no transition pathway existed.

As artificial intelligence changes work, the right to retraining is becoming an attractive policy promise. It is more constructive than treating displacement as an unavoidable side effect of innovation. It recognises that workers should have a fair opportunity to adapt when employers and economies benefit from new technology.

But access to a course is not access to a future.

A meaningful right to retraining must include the conditions under which a person can learn, practise, demonstrate new capability and enter changed work. Without those conditions, retraining becomes a transfer of responsibility. The institution changes the job; the individual is left to solve the consequences.

Training Is an Input, Not a Transition

Training is easy to announce because it is easy to count.

Organisations can report licences issued, modules completed, badges awarded and hours of learning consumed. Governments can report people enrolled in national skills programmes. Technology companies can report scholarships and course access.

These numbers matter, but they do not establish that a worker moved from a declining role into a viable one.

A transition requires more than knowledge acquisition. The person must understand which role they are moving towards. They need time to learn. They need access to the relevant tools and context. They need practice on work that resembles the destination role. They need feedback. Their competence must be recognised by somebody with the authority to open the next door.

The transition is complete only when the person's position in the labour market has changed.

This distinction is especially important in South Africa. In a labour market with high unemployment, workers cannot assume that a newly learned skill will meet abundant demand. An employed person who leaves work to study may carry substantial risk. An unemployed person may complete several programmes without acquiring the experience employers require. Training can improve capability while opportunity remains constrained.

The right to retraining should therefore not be confused with a right to a guaranteed job. No institution can promise that every transition will succeed. It should, however, require an honest and adequately supported route from current work to plausible future work.

The People Most at Risk Often Have the Least Capacity to Learn

The workers most exposed to task change are not necessarily the people best positioned to retrain.

They may have less control over their schedules, weaker digital access, limited confidence after years away from formal study, family responsibilities or transport constraints. They may be working under pressure in the very process being redesigned. Some will be asked to learn the new system while continuing to deliver the old one.

An invitation to learn after hours treats time as if it were equally available to everyone.

It is not.

The OECD's work on adult learning shows that participation is stagnating or declining in many countries and that much workplace training remains short and compliance-driven. The people who most need substantial development often face the greatest barriers to participating in it.

If retraining matters to the organisation's operating model, it should occur partly inside paid work. Protected learning time is not an employee benefit added around the edges. It is an investment required to produce the future workforce on which the new model depends.

Reskilling Without Job Redesign Is Theatre

Many reskilling programmes begin with a catalogue of courses rather than an examination of work.

Employees are offered AI literacy, data analytics, cybersecurity, project management or coding. These are all potentially useful. But the organisation may not have identified which roles will change, which tasks will move to machines, which new responsibilities will remain human or how many opportunities will exist.

The worker is asked to choose a future from a menu before the organisation has designed one.

Real retraining starts with the operating model.

Which work is disappearing? Which work is growing? Which existing jobs will be augmented rather than removed? Which capabilities are adjacent enough for people to cross into them? Which roles can be decomposed so that a learner can contribute before reaching full proficiency? Who will supervise the transition?

Only then should the curriculum be designed.

A call-centre employee may not need a generic course in artificial intelligence. They may need to move from answering routine queries to inspecting difficult cases, correcting automated classifications, identifying recurring failure patterns and improving the knowledge base. That transition requires domain knowledge, judgement, customer skill, data interpretation and a clear escalation mandate.

The course is only one component. The job must be rebuilt around the capability.

Practice Is the Missing Right

People do not become capable because they have been told how new work operates. They become capable by doing it under conditions that allow error, feedback and increasing responsibility.

This creates a practical problem. Employers often want reskilled workers to prove themselves before being moved into the work, while workers need access to the work in order to become competent.

The organisation must create a protected bridge.

That bridge may include simulations, shadowing, rotations, supervised projects, temporary assignments, apprenticeships or partial responsibility for live work. The learner should move from observation to assisted performance, then to reviewed independence.

This is not fundamentally different from how professions have always developed people. What changes is the speed and breadth with which these pathways now need to be created.

A serious right to retraining should therefore include a right to relevant practice. Without practice, the worker receives knowledge but not employability.

Evidence Must Be Recognised

Retraining also fails when the evidence it produces has no authority.

An employee completes a programme, but the internal recruitment system still filters by previous job title. A supervisor agrees that the person is capable, but the formal role requires an external credential. A digital badge records attendance, not performance. A worker moves companies and discovers that the evidence is not portable.

This turns development into a private learning experience rather than a labour-market asset.

Every retraining pathway should specify what competence will be assessed, who recognises it and which opportunities it unlocks. The evidence might include an observed demonstration, a reviewed project, a portfolio of decisions, a formal assessment or a period of successful supervised work.

The purpose is not to create more certificates. It is to make the transition legible to the people controlling access to work.

Managers, human-resources systems, professional bodies and hiring processes must recognise the same evidence. Otherwise the employee reaches the end of the programme and finds that the institutional door has not moved.

The Employer Cannot Outsource the Whole Obligation

Workers have responsibilities. They need to participate, practise, accept feedback and remain willing to learn. Governments must support education systems, employment services and social protection. Training providers must produce relevant, credible learning.

Employers still carry a distinctive obligation because they decide how technology is introduced into work.

They select the systems. They capture much of the productivity benefit. They redesign roles, remove tasks, change performance expectations and determine whether internal mobility is possible. They have access to the work context and the supervisors through which capability can develop.

An employer cannot make a consequential design choice and then treat the affected worker's adaptation as an entirely private project.

This does not mean every employer must preserve every job. It means that a claim of responsible transition should be tested against more than the availability of training.

Did the organisation give people notice? Did it identify plausible destination roles? Did it provide paid time, tools and practice? Did it adjust selection rules? Did it create real opportunities for internal candidates? Did it track what happened to participants after completion?

If not, the retraining promise may be reputational cover for ordinary displacement.

Managers Are the Conversion Point

Retraining strategies often sit with human resources or learning and development. The actual conversion happens through managers.

Managers decide whether employees receive time away from current work. They assign practice opportunities. They tolerate the initial drop in productivity that comes with learning. They provide feedback, recommend people for roles and decide when somebody can be trusted with greater responsibility.

A programme can be excellent and still fail because line managers are rewarded only for short-term delivery.

This creates a familiar contradiction. The organisation says future capability is strategic, while the manager is punished if present output slows. The manager therefore releases the employee for an online module but not for meaningful practice. Completion rises. Capability does not.

If retraining is strategic, managers need explicit capacity, targets and authority to develop people through the transition. The cost must appear in the plan. Learning time must be treated as planned capacity, not absence. Supervising a transition must count as management work.

What a Meaningful Right Would Contain

A workable right to retraining need not promise unlimited education or permanent employment. It should establish a minimum transition architecture.

At least eight elements matter:

1. Advance visibility. Workers receive credible information about how their work is likely to change.
2. A defined destination. Training connects to identifiable roles, tasks or income opportunities.
3. Protected time. A meaningful portion of learning occurs within paid work.
4. Access to tools. Workers can practise with the systems and data relevant to the new work.
5. Supervised practice. Learning includes performance, feedback and increasing complexity.
6. Recognised evidence. Completion produces proof that hiring managers and institutions accept.
7. A fair opportunity. Internal candidates receive real consideration for redesigned work.
8. Outcome tracking. The organisation measures transition, retention, progression and earnings-not only course completion.

These conditions do not guarantee success. They make the offer credible.

From Learning Benefit to Transition Institution

Artificial intelligence will alter different occupations at different speeds. Exposure to AI is not the same as job loss, and much of the change will occur through task transformation rather than whole-job disappearance. That makes good transition design more achievable, not less necessary.

Workers can move while parts of their existing knowledge still have value. A domain expert can become the person who reviews automated work. A frontline employee can become an exception handler. An administrator can become a workflow controller. A technician can use intelligent tools to diagnose more complex problems.

But those transitions will not happen because a course exists.

They require institutions that convert existing experience into new capability, create room for practice and give credible evidence access to opportunity.

The right to retraining is a good starting principle. It says people should not carry the full cost of technological change alone.

It is not enough.

The real right is a fair chance to make the transition: time to learn, work through which to practise, evidence that can be trusted, and a door that genuinely opens when the capability has been demonstrated.

Anything less risks turning reskilling into a polite description of being left behind.

Sources

• OECD, AI and Skills (2026)
• OECD, Trends in Adult Learning (2025)
• International Labour Organization, Generative AI and Jobs: A Refined Global Index of Occupational Exposure (2025)
• International Labour Organization, Changing Landscape of Skills in the Age of AI (2026)
• World Economic Forum, Artificial Intelligence and the Future of Entry-Level Work (2026)