Can You Hire Your Way Out of an AI Skills Gap?

Imagine two meetings taking place in the same organisation. In one, leaders approve a search for people with artificial intelligence expertise. In the other, an employee asks for time and training to learn how AI could improve their work and is told the budget is still being considered.

Both conversations might make sense on their own. Together, they reveal an interesting problem.

The organisation may be searching outside for a capability it has made difficult to develop inside.

And as demand for AI skills grows, that tension is becoming harder to ignore.

Recent UK research from Robert Half found that 47 per cent of technology and IT employers planned to increase permanent hiring before the end of 2026. Cybersecurity, agentic AI, generative AI and cloud were among the capabilities attracting the greatest demand.

Meanwhile, PwC’s 2026 global workforce research found something happening in the opposite direction. Only 51 per cent of workers said they had access to the learning and development resources they needed, down from 59 per cent a year earlier.

So employers need new skills. Employees need new skills. But are we too quick to assume those skills need to arrive through the front door?

A Skills Gap Isn’t Necessarily a Recruitment Problem

The obvious response to a skills shortage is to recruit. Sometimes that’s exactly right. A business may need expertise that would take years to develop internally. It might need additional capacity for urgent work. Or it may benefit from bringing in someone who can challenge established thinking.

The problem starts when recruitment becomes the whole skills strategy. “We need AI skills” sounds specific. It isn’t. One team may need an engineer capable of building and governing AI systems. Another may need managers who can redesign workflows around them. Another may simply need employees who know how to use an approved AI tool, question its output and recognise when human judgement needs to take over.

Those are very different capability gaps. Put them all into a single job description and you can easily end up searching for a mythical candidate who already possesses everything. Then, when you can’t find them, you conclude there’s a skills shortage. Perhaps there is. But perhaps the first question should be:

What capability do we actually need?

Break the Gap Apart

Before opening another vacancy, look at the work. What outcomes are you trying to improve? Where could AI genuinely help? What knowledge of your customers, organisation, regulation or industry is essential? Which capabilities require genuine specialist expertise? And which could someone already in the organisation learn through guided practice?

Those questions may produce a very different answer. You might need to hire. You might need to develop existing people. You might need to redesign the role. Most likely, you’ll need some combination of all three.

That’s a much more useful workforce strategy than simply adding “AI experience required” to another job advert.

Look at Who Has Actually Had the Chance to Learn

There’s another problem with recruiting only for demonstrated AI experience.

Not everybody has had the same opportunity to acquire it.

PwC describes 56 per cent of the global workforce as the “engine room”: people responsible for much of the everyday work within organisations, whose skills are less scarce and who have so far gained less advantage from AI.

Fewer than two in five of this group said they had access to the learning and development resources they needed. That matters. Because current capability is partly a record of opportunity.

Give someone access to good tools, interesting problems, supportive managers and time to experiment and they’ll begin accumulating evidence of AI experience. Give somebody else none of those things and, a year later, they may appear to be the weaker candidate. That doesn’t necessarily tell us which person had greater potential. It tells us who had the opportunity to develop it.

Training Isn’t Enough Either

There’s a temptation at this point to swing too far in the other direction. Fine. We’ll train everyone. But buying an online AI course isn’t a skills strategy either. People learn by applying knowledge. Useful development needs to connect to real work. Give someone an approved tool. Give them an actual problem. Give them clear safeguards. Give them access to someone who understands the task. Give them feedback. And, importantly, give them enough time to practise.

They also need permission to challenge the technology. AI capability isn’t simply knowing which button to press or how to write a clever prompt. It includes knowing when the output is wrong. When it needs checking. When it creates risk. And when human judgement should override it.

Interestingly, Robert Half found that 45 per cent of UK technology professionals are already expected to develop new AI-related skills, while 38 per cent are spending more time overseeing and validating AI-generated output.

The job is already changing. Development needs to keep up.

Then Recruit for the Gap That Remains

Once you’ve defined the work properly and explored the capability already inside the organisation, recruitment becomes much more precise.

Now you can distinguish between:

What somebody genuinely needs to know on day one.

And:

What a capable person could reasonably learn once they’re here.

That’s an important distinction. It can widen the candidate pool without lowering the standard. Someone who has worked with a different AI platform may still have exactly the analytical habits you need. Someone from an adjacent discipline may understand your customers exceptionally well and learn the technology quickly. Someone without the perfect list of AI keywords on their CV may have excellent judgement, curiosity and learning agility. And when specialist expertise genuinely is required, hire it. Bring in the engineer. Recruit the expert. Use the contractor. But perhaps make transferring some of that knowledge part of their job. Then the organisation is stronger when the project ends than it was when it began.

Don’t Create an AI Divide Inside Your Own Organisation

Morgan McKinley’s 2026 global AI research highlights the same tension. It found that 52 per cent of employers planned to increase AI training budgets, yet 58 per cent of employees reported receiving inadequate or no AI training. Its employer sample was relatively small, so the exact percentages deserve some caution, but the underlying question is still worth asking.

Are the people who most need the opportunity to develop actually getting it?

Because there’s a danger of creating two workforces. One group gets access to the technology, training and interesting projects. Their skills accelerate. The other keeps doing the existing work. Their opportunity to develop shrinks. Then we look at the difference between those two groups and call it a skills gap. Some of it undoubtedly is. Some of it may be an opportunity gap we’ve created ourselves.

Hiring and Development Should Feed Each Other

This isn’t an argument against recruitment. New colleagues bring expertise, different experiences, challenge and momentum. Existing employees bring something equally valuable: context, relationships, organisational knowledge and an understanding of how the work really gets done. You need both.

The more interesting question is one of sequence.

• Define the work.
• Understand the capability you already have.
• Give people a fair opportunity to develop the skills that can be learned.
• Redesign roles where old boundaries no longer make sense.
• Then recruit for the expertise and capacity that genuinely remain missing.

That approach is better for candidates too. A focused role is easier to understand. A realistic assessment gives someone a better opportunity to demonstrate what they can actually do. And a credible development path makes the opportunity more attractive after they join. Most importantly, it acknowledges something fundamental about AI.

Nobody is arriving finished.

The technology is changing too quickly. The AI expert you recruit today will need to keep learning tomorrow, just like the people already working for you. An organisation that can only acquire new capability from the market will therefore keep returning to the same problem.

Hiring can absolutely help close an AI skills gap. But the strongest organisations won’t just buy the skills they need. They’ll get better at creating them too.