The AI Journal

A certificate is a persuasive object because it compresses a complicated story into a simple signal. It says that a person enrolled, persisted, met stated requirements and reached the end of a course. Employers, regulators and learners can understand it quickly. The problem is that completion is routinely asked to carry a meaning it cannot support: that the learner is now capable of performing the relevant work.

Finishing a course demonstrates completion. Competence requires performance. Those statements are not equivalent.

This distinction matters most in short professional courses, where the gap between learning and work is often concealed by the apparent neatness of the programme. A learner may complete a six-week course in data analysis, validation, quality systems or instructional design by watching lessons, passing quizzes and submitting a final assignment. The provider can report a completion rate. The learner can display a credential. Yet neither fact establishes that the learner can select an appropriate method when the data are incomplete, document a defensible decision, challenge an unsafe assumption or adapt a procedure to an unfamiliar case.

The classic literature on training transfer defines transfer as the effective application of knowledge, skills and attitudes from the training context to the job, including both generalisation and maintenance over time (Baldwin and Ford, 1988). This definition exposes what completion leaves out. A course can produce learning inside the course while failing to produce changed behaviour outside it. Transfer is not an automatic final stage that appears once a learner has accumulated enough instructional hours. It depends on the design of the learning experience, the learner’s characteristics and the environment in which performance is expected.

The distinction is not merely philosophical. Blume and colleagues’ meta-analysis of 89 empirical studies found that transfer was associated with factors including cognitive ability, conscientiousness, motivation and a supportive work environment. The relationships were also affected by the type of skill being trained and by how transfer was measured (Blume et al., 2010). This makes completion an especially poor headline outcome. It attributes success to the course while ignoring the conditions that allow a learner to use what the course taught.

A learner who completes a programme but has no opportunity to practise the target behaviour has not necessarily failed. The organisation may have given them no time, tools, authority or feedback. A laboratory technician may understand a new digital system but never be permitted to use it. A supervisor may know the principles of root-cause analysis but be required to close investigations at a speed that makes careful analysis impossible. A lecturer may learn a new assessment method but work within a timetable that leaves no space for redesign. In each case, a completion statistic hides the difference between personal learning and organisational enablement.

The usual response is to add more content. If learners are not competent, the reasoning goes, the course should contain another module, another reading or another quiz. But additional exposure is not the same as additional capability. Training research treats design, delivery and implementation as a connected system; what happens before and after instruction can matter as much as what happens during it (Salas et al., 2012). The answer to weak transfer is therefore not necessarily a longer course. It may be a better task, a more realistic constraint, a supervisor conversation, a period of supported practice or a later assessment conducted in the environment where the skill matters.

This suggests a different definition of a successful short course. Its central question should not be, “How many learners reached the end?” It should be, “What can a learner now do, independently and reliably, that they could not do before?” The question is harder because it demands evidence. It also produces more useful evidence.

The most defensible evidence will usually be performance-based. Instead of asking learners to recall the steps in a deviation investigation, a programme might give them a messy scenario with conflicting records and require them to identify the problem, justify the investigation pathway and produce a concise report. Instead of asking whether learners understand data visualisation, it might require them to choose a suitable chart, explain its limitations and make a decision from imperfect data. Instead of testing whether a learner can describe a process, it might ask them to improve one while preserving quality, safety and traceability.

Such assessments need not imitate the whole workplace. A carefully designed simulation, case, portfolio or observed demonstration can provide useful evidence without pretending to measure every dimension of professional competence. The important feature is that the task requires judgement and production, not only recognition. Authentic assessment is valuable precisely because it evaluates whether learners can use knowledge in a context that resembles the demands for which the course claims to prepare them (Wiggins, 1998).

This does not make completion irrelevant. Completion remains useful for administration, learner motivation and progression. A provider needs to know who finished, and a learner may reasonably want recognition for sustained effort. Completion can also be an early indicator that a programme is accessible and that its workload is manageable. But it should be reported as what it is: evidence of participation through a defined learning experience. It should not quietly become evidence of workplace competence.

The strongest objection to this argument is practical and forceful: transfer is expensive and sometimes impossible to measure. Course providers may not control the learner’s workplace. Employers may regard post-course observation as intrusive or confidential. Learners may be studying for future roles and have no relevant setting in which to demonstrate performance. A demand for transfer evidence could disadvantage precisely the people short courses are meant to support, while allowing affluent learners with better workplace access to appear more competent.

That objection narrows, but does not defeat, the claim. Transfer should be the primary design target, not an inflexible requirement that every provider observe every learner on the job. Where workplace observation is impossible, programmes can use structured simulations, role-specific projects, delayed demonstrations, supervisor attestations or portfolios of applied work. They can distinguish “demonstrated in simulation” from “demonstrated in practice” rather than collapsing both into the same certificate. The point is not to create a perfect measure. It is to stop treating no measure as if it were one.

There is also a fairness argument for this distinction. When certificates imply competence that has not been demonstrated, the burden falls on learners who trust the credential and employers who must interpret it. A transparent statement of evidence gives both parties more agency. It allows a learner to say, “I completed the theory and demonstrated the process in a simulated case,” which is more honest and more informative than a certificate that merely announces completion. It also encourages employers to provide the missing conditions for transfer instead of assuming that education has already solved the performance problem.

Programme quality processes should therefore separate at least three outcomes: participation, learning and application. Participation asks whether the learner completed the course. Learning asks what the learner knows or can perform under assessment conditions. Application asks whether that capability is used, retained and adapted beyond those conditions. The three outcomes are related, but they are not interchangeable, and a credible quality system should resist presenting them as a single ladder in which one automatically proves the next.

The future of professional education will not be secured by issuing more certificates. It will be secured by making the claims attached to certificates more precise. Completion deserves recognition, but competence deserves evidence. A short course becomes more valuable—not less accessible—when it tells learners and employers exactly what has been demonstrated, what remains to be practised and what conditions are needed for learning to become performance.