AI-Rich, Experience-Poor: Procurement's Next Capability Gap

AI is making it possible for an early-career professional to produce work that once required far more experience. A market summary can be assembled in minutes. Supplier arguments can be anticipated. A negotiation plan can be challenged before the meeting. Drafts become cleaner, faster and more complete.

That is progress. It also changes how experience is accumulated.

Pearson's new For Every Future research, launched on 24 September 2026, describes a "triple capability gap" across skilled trade, technical and service occupations. As AI changes those jobs faster than training systems adapt, experienced workers are also retiring and taking practical knowledge with them. Pearson warns that early-career talent risks becoming rich in AI support while remaining poor in lived experience.

The report did not study procurement. It is useful here as an adjacent workforce signal, not as evidence about buyers specifically. The mechanism it describes is nevertheless worth examining because procurement is also a profession in which judgment is built partly through repeated exposure to situations that do not have a clean answer.

A separate 2026 ProcureAbility benchmark makes the capability question directly relevant to procurement. Among 160 senior procurement leaders, 71 percent described their teams as moderately or highly capacity constrained. Only 19 percent were adding headcount, while 46 percent were investing in upskilling existing people. That does not validate Pearson's mechanism in procurement, but it does show why the design of internal capability building matters now.

Some inefficient work was also practice

There is no reason to preserve a slow task simply because somebody once learned from doing it manually.

A junior buyer does not become better because they spend three hours formatting a supplier comparison that AI can structure in five minutes. Nor is there virtue in rewriting the same meeting notes, searching documents by hand or producing a first draft that a model can generate more accurately.

The problem appears when the organisation removes the task and assumes it has removed only the waste.

Routine work often carried secondary learning. Reading ten supplier proposals exposed someone to ten different commercial structures. Sitting beside a category manager while they prepared a negotiation revealed which facts mattered and which were noise. Drafting a response forced the junior person to choose a position before seeing the senior person's answer. Joining a lower-stakes supplier call created a safe place to hear resistance, miss a cue, recover and remember what happened.

AI can compress much of the output work around those activities. It does not automatically replace the experience embedded inside them.

That concern sits beside, but is different from, the question in our PISA 2026 article. There the issue is whether AI can reduce the cognitive effort through which something is learned. Here the issue is exposure: whether the work system still gives people enough situations in which practical judgment can form.

Procurement has an apprenticeship problem

Commercial judgment is difficult to teach as a list of rules because the right move depends on context.

A supplier asks for an eight percent increase. The buyer may need to challenge it hard, trade something, delay the discussion, escalate internally or accept part of it. The answer depends on margin structure, alternatives, supply risk, timing, history, the credibility of the claim and what else is moving in the deal.

Experienced procurement professionals recognise those patterns because they have seen versions of them before. They have watched a weak threat collapse, heard a real constraint, seen a stakeholder overstate urgency and learned what a supplier sounds like when there is genuinely no room left.

If AI lets junior professionals arrive with better preparation earlier in their careers, that is valuable. But the organisation still has to create the encounters through which they learn what the preparation means when another person resists it.

This becomes more important as senior experts become busier or leave. Tacit knowledge does not transfer through a document repository alone. Some of it only becomes visible when somebody explains why they ignored one signal, pursued another or changed course halfway through a conversation.

The entry-level work is changing before development systems catch up

Fresh German workforce data makes the problem less theoretical.

A representative IU International University study published on 24 September 2026 surveyed 2,000 employees aged 16 to 65. The sample was weighted to represent the German labour market by age and gender.

Among respondents, 41.3 percent agreed somewhat or completely that AI is taking over tasks previously performed by career entrants. At the same time, 43.9 percent had spent no working time at all on AI training during the previous 12 months. Only 33.6 percent said their employer supported the development of critical thinking strongly or fairly strongly.

That study is not about procurement, and these figures describe employee perceptions rather than measured skill loss. But they strengthen the mechanism behind the capability question: organisations may remove traditional entry-level work faster than they redesign how early-career people gain judgment.

For procurement, the risk is not that a junior buyer no longer formats a supplier comparison manually. The risk is that the work disappears together with the observation, challenge, discussion and repetition that used to surround it, while no deliberate substitute is created.

The answer is still not to preserve inefficient tasks. It is to become explicit about which developmental repetitions were hidden inside them and rebuild those repetitions somewhere better.

Do not preserve the old work. Replace the lost repetitions.

The answer is not to slow AI adoption so junior people can keep doing administrative work for developmental reasons.

It is to separate productivity design from capability design.

If AI removes two hours of manual preparation, part of that time can be returned as structured exposure: observing a difficult supplier meeting with a clear learning objective, comparing the AI recommendation with the senior buyer's decision, running a short after-action review, taking ownership of one part of a live conversation, or rehearsing the same situation before the stakes are real.

The distinction matters because passive observation alone is not enough. People need chances to make a decision, say the words, receive feedback and try again. That is also the argument behind our piece on AI and workplace conversation skills: as technology handles more of the surrounding work, the remaining human moments may become fewer but more demanding.

For procurement leaders, the practical question is therefore not whether junior buyers should use AI. They should. The question is what replaces the experience that disappears when the old workflow becomes faster.

A function that automates routine work without redesigning how people gain judgment may become more productive while weakening its future bench. A function that uses the saved time for observation, responsibility, feedback and realistic spoken practice can get both benefits: better tools now and stronger negotiators later.