Part of professional education and recruitment has been organised around operational predictability. Universities and businesses have taught and assessed technical sequences, analysis, documentation and calculation. That learning remains valuable; the problem arises when fast, uniform execution becomes the dominant indicator of competence and promotion.
The spread of generative artificial intelligence changes that design. It can accelerate report synthesis, programming and certain stages of design, although quality and reliability vary by task. Making part of execution cheaper requires reconsidering the experience people need to interpret results and take responsibility for decisions.
Changes in recruitment and education invite us to examine an institutional mismatch. They do not prove that the entire previous market has disappeared or that every employment difficulty has a single cause: they raise the question of whether we still prepare and select people using criteria inadequate for the work they will need to perform.
The disruption of the operational foundation
Stanford HAI’s AI Index 2026 puts the decline in employment among software developers aged 22 to 25 in the United States at nearly 20% since 2024, rather than necessarily over twelve months. Its education chapter records an 11% reduction in undergraduate computer science enrolment at US four-year universities between 2024 and 2025, rather than across all technical faculties. These indicators cover different areas and do not, on their own, establish AI as the sole cause. Meanwhile, The State of Fashion 2026, by The Business of Fashion and McKinsey, describes workforce reorganisation and the need for reskilling.
From that context, we can propose an editorial risk for fashion: if an organisation eliminates opportunities for early practical experience without offering alternatives, it may weaken the environment in which future leaders develop intuition about fabric drape, stitching and volume. This application is not presented as a literal conclusion by BoF’s analysts.
Replacing part of early-career work with software can improve immediate productivity. The balance changes if the transmission of technical judgement is interrupted without rebuilding mentoring, practice and career progression. Savings and learning must be assessed together.
The paradox of selection bias
As entry-level roles contract, it is worth preserving capabilities that complement assistance: proven experience, problem framing and responsibility in decision-making. These capabilities can be found in diverse careers; neither age nor mastery of a tool establishes professional soundness on its own.
Yet ordinary recruitment processes continue to operate with parameters inherited from the previous cycle:
1. The algorithmic penalty on experience
Screening tools and recruitment practices can make experienced careers invisible when they privilege keywords about recent tools or incorporate assumptions about age and adaptability. Familiarity with an interface is not equivalent to leadership competence. Experience can contribute knowledge of balance sheets, teams and processes, but must be assessed through evidence relevant to the role, without assuming that everyone over fifty has superior judgement or that every recruitment system penalises them.
2. The illusion of technical self-sufficiency
A junior operator equipped with advanced software does not automatically equal an experienced executive. Nor do they necessarily lack judgement. The AI Index 2026 reports productivity improvements alongside negative or limited results in other settings: the effect depends on the task, the tool and its use. Assisted speed must be accompanied by understanding, review and responsibility proportionate to the risks.
3. The multiplier from absorbed administrative work
One opportunity offered by automation is to expand the capacity of professionals with demonstrated judgement, both experienced and developing. Absorbing part of data processing, draft preparation and coordination can free time for complex diagnoses and relationships with specialist suppliers. This expansion is not automatic: quality, supervision costs and continuity of learning must be measured.
The cost of assessing tools instead of judgement
Responsibility for this gap does not lie with technology or with people trying to enter the labour market. It lies with the design of corporate governance systems and the metrics used to assess talent.
As long as boards and faculties measure performance by the speed at which documents are produced or software is applied, the system will continue to reward operational conformity and exclude individuality.
The educational and institutional challenge extends beyond teaching applications that change rapidly. It also involves redefining the skills that should be cultivated as certain execution tasks become cheaper:
- Teach people to formulate problems and understand solutions, alongside competent use of tools.
- Rebuild practical spaces where young talent can make mistakes with real materials before delegating to virtual models.
- Review recruitment processes to recognise demonstrated judgement, relevant experience and learning ability without making age a substitute for assessment.
