GUEST COLUMN:
Dr. Gleb Tsipursky
CEO
Disaster Avoidance Experts
Wales has no shortage of discussion about AI skills. The harder question is what happens after someone has the course, qualification or apprenticeship place and arrives in an AI-enabled workplace.
If the technology performs much of the routine work that beginners used to do, employers need a new design for how people become experienced.
The Stanford Digital Economy Lab’s August revision gives that problem urgency. The employment shortfall for workers ages 22–25 in highly AI-exposed occupations widened from 15% in the July 2025 data vintage to 19% by June 2026. The adjustment appears mainly through reduced hiring of young workers and is concentrated where AI use tends to automate human tasks.
Business News Wales has recently argued that Wales needs to train for the economy it is building and that digital apprenticeships can hold the key to successful AI adoption. The next step is to redesign the work apprentices actually do.
Consider advanced manufacturing, semiconductors, cyber security, professional services and digital industries. AI can generate first-pass analysis, compare technical information, draft customer material and surface anomalies. Those capabilities should not make apprentices peripheral. They should let apprentices get to the learning-rich part of the job sooner.
A manufacturing trainee can spend less time compiling a report and more time investigating the variance that could stop a line. A junior cyber professional can move from routine log review into supervised analysis of unusual patterns. A new finance employee can let AI prepare the reconciliation and focus on the transactions that do not fit. A developing digital professional can test the assumptions in an AI-generated plan rather than spend hours formatting the plan.
The principle is simple: automate preparation, preserve ownership of the judgment.
Employers then need managers who coach that judgment. Before signing off an AI-assisted recommendation, ask the learner what they verified, which exception matters, what evidence would change their view and when they would escalate. Those questions turn live work into deliberate practice.
Wales can also measure apprenticeship quality differently. Completion matters, but so does time to independent competence. How quickly can a learner handle a defined set of real decisions without rescue? Does AI make that journey shorter?
That metric aligns technology adoption with economic development. It rewards firms not simply for using AI, but for becoming better at turning new entrants into capable professionals.
This is particularly important in a country trying to keep talented young people connected to emerging opportunities. If employers automate junior work and then demand experienced hires, they can worsen the very skills shortage they are trying to solve.
Wales already has a strong apprenticeship infrastructure and a growing set of digital and AI initiatives. The opportunity is to combine them at the workflow level.
AI should remove low-value repetition. Apprenticeships should provide more high-value repetition: verification, exceptions, communication, problem-solving and supervised decisions.
That is how Wales can build both a more productive economy and a deeper pipeline of people ready to lead it.
