Become a TechRadar Insider
The technology sector has a habit of spotting contradictions everywhere except in its own workforce strategies. Today’s is particularly stark.
Globally, 74% of employers struggle to find qualified talent, with $11.5 trillion in annual productivity lost to skills gaps.
Yet at the same time, overall tech hiring remains materially below pre‑pandemic levels, with entry‑level roles contracting far faster than the rest of the market as AI absorbs routine work.
Article continues belowVP at Cornerstone.
Nowhere is that tension clearer than in AI. There are around 1.6 million unfilled AI roles worldwide, even as the early‑career jobs that once allowed people to build those skills are quietly disappearing.
Employers are calling for advanced capability while narrowing one of the main routes through which future expertise is created.
AI isn’t creating “AI jobs” – it’s reshaping every job
Demand reflects that shift. Our data found that AI and machine learning skills have grown 245%, making them one of the fastest‑growing technical skill categories. At the same time, the traditional split between “technical” and “human” work has collapsed; roles now require a near‑even mix of both.
These capabilities do not appear fully formed. They are developed through exposure to real work, judgement calls and context - historically gained through early‑career roles. Increasingly, those are precisely the activities being absorbed by AI tools.
The impact of Entry‑level role exposure
Entry‑level roles are especially exposed because so much junior work is structured and repeatable. Tech job adverts have declined 50% since 2019/20, with junior roles among the hardest hit. Projections suggest a 45% decline in junior developer roles, alongside steeper falls in AI QA testing and basic IT support.
Both external research and our own data show organizations can already automate around 30% of entry‑level work hours, including tasks that once acted as informal apprenticeships into more complex roles. On paper, this looks like progress as productivity improves, what’s harder to see is the delayed cost.
When fewer juniors are hired, and those who encounter thinner, more automated roles, fewer people accumulate the experience needed to step into senior positions later. A few years on, employers find themselves searching for advanced AI capability in a market that has quietly produced too little of it.
This is the structural problem beneath the AI talent shortage that we’re finding - companies are accelerating output today while eroding the learning capacity their future workforce depends on.
AI is reshaping jobs as we know them
AI is not eliminating jobs wholesale, it’s reshaping the work inside them. Our analysis shows job titles remain stable while skills within them change rapidly - a form of substitution that headcount figures alone fail to capture.
For example, architects worried when AutoCAD architecture software arrived - manual drafting disappeared, but the profession strengthened. AI has a similar effect in technology - used well, it can remove friction and free up time so people can focus on the more human elements of the role.
The mistake is treating productivity as the only outcome that matters. Short‑term efficiency gains can coexist with long‑term capability loss, and leaders rarely see the latter until it becomes expensive.
From AI literacy to AI excellence
A more sustainable approach treats AI adoption as a maturity journey, not a switch.
It starts with AI literacy - understanding what tools can and cannot do, where they add value and where they introduce risk. Literacy is no longer optional and needs to be encouraged, but it’s only the first step.
From there comes fluency, where AI is used inside real workflows and decisions, with humans still accountable for outcomes. Finally comes excellence, where AI is embedded into operating models, governed properly and continuously improved.
Early‑career roles matter at every stage of that journey. The task now is to protect the learning value inside work. That does not mean preserving routine tasks, but rather identifying which activities still matter because they help less‑experienced employees understand the very foundations of their roles.
In many cases, AI should sit alongside junior staff, with explanation, review and correction built into the process so that learning continues rather than quietly disappearing.
What tech leaders should do next
If leaders want the AI dividend to last, workforce design needs as much attention as tooling. Automating tasks without redesigning roles is an incomplete strategy, particularly at the entry level where capability is formed.
Rebuilding junior work starts with deliberately protecting its learning value, even when AI is used to assist. Where automation is introduced, explanation and review need to be built into the workflow so that judgement continues to develop.
Training must also move beyond basic awareness. Organizations should support progression from literacy into fluency, where AI is used inside real decisions and workflows, and ultimately into excellence, where human‑AI collaboration is embedded into how work gets done.
This is also where skills strategy needs to mature. In a market shaped by constant change, skills should be treated like R&D rather than a compliance exercise. Leaders need clearer visibility into which capabilities are rising, which are fading and where adjacent skills already exist inside the organization.
Workforce intelligence belongs at the center of growth planning, with success measured by whether capability is expanding over time, not just whether output ticks up in the next quarter.
Reskilling pathways and internal mobility also need to be normalized rather than treated as an exception. Automation will continue to remove parts of roles, particularly at the execution layer.
What matters is whether people have credible routes into the work that replaces them. Redundancies may be a short‑term outcome of rapid change, but the long‑term requirement is reskilling.
The warning – and the opportunity
AI will continue to raise expectations of speed and sophistication in technology work, but capability still has to be grown.
We must invest in early‑career development, redesign the work that teaches people, and treat skills as the growth asset they are.
AI can accelerate technology, but only if leaders stop eating their young.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit