Most recruitment teams have felt it this year. A role opens that asks for “strong AI skills”, the shortlist looks promising, and then the interviews show that very few candidates can explain how they use AI tools in real work. The market is short on people who are confident with AI, and hiring managers are competing for the same small group.
For many employers, the answer is not to keep searching. It is to build those skills inside the team they already have. That shift changes the work of HR, learning teams and recruiters, and it is worth looking at how it plays out in practice.
The gap is wider than job adverts suggest
AI tools are already in daily use across most offices. Staff use them to draft emails, summarise reports, check data and prepare for meetings. But using a tool is not the same as using it well. Many employees learned by trial and error, with no guidance on what is safe to share, what good output looks like, or when to trust the result.
The skills gap shows up in other ways too. Skillsoft research reported by Onrec found that more than half of managers say staff are assigned work beyond their skills, and linked that gap to slower AI adoption. The World Economic Forum has also estimated that most of the global workforce will need some form of reskilling by 2030. For HR leaders, that means the AI question is no longer about a few technical roles. It touches almost everyone.
Why recruiting your way out rarely works
When a skill is scarce, the first instinct is to hire for it. With AI, that approach runs into three problems.
The first is cost. Candidates who can show real results with AI tools know their value, and salary expectations reflect it. Paying a premium for every new hire adds up quickly, and it can create tension with existing staff doing similar work.
The second is speed. AI tools change every few months. A candidate hired for their skills in one tool may need retraining within a year anyway. Hiring solves today’s gap but does not keep the team current.
The third is context. Even a skilled new hire does not know your processes, your data rules or your clients. Some of the most useful AI work happens when someone who knows the business well learns how to apply the tools to it. That person is often already on your payroll.
None of this means employers should stop hiring for AI skills. It means hiring works best alongside a plan to develop the people already in post.
What good in-house AI training looks like
The weakest approach is the one many companies try first: a one-off webinar, a shared folder of prompts, and a note encouraging staff to “experiment”. It feels like progress, but usage tends to drop off within weeks, and the gap between confident and unsure employees grows.
Stronger programmes share a few features.
They start with roles, not tools. A finance analyst, a recruiter and a sales manager use AI in very different ways. Training that shows each group how the tools fit their own daily tasks gets far more uptake than a general overview.
They are live and hands-on. People learn AI by doing, with someone on hand to answer questions and correct mistakes. Recorded videos have a place, but most employees need practice on real tasks to build confidence.
They cover safe use. Staff need clear rules on what data they can put into which tools, how to check outputs for errors, and when human review is required. This protects the business and gives cautious employees permission to start.
They include managers. Teams follow their manager’s lead. If managers do not understand how their people use AI, they cannot set realistic targets or spot where the tools add value. Some of the most effective programmes train leaders first.
For companies without the capacity to design this in-house, specialist providers now offer structured ai training for employees built around live classes tailored to each job function, which takes much of the design work off the learning team. Whichever route an employer chooses, the aim is the same: practical skills that people use the following week, not a certificate that sits in a file.
Measuring whether it worked
Learning teams have long been judged on attendance and completion rates. With AI training, those numbers say very little. A team can complete every session and still not change how it works.
The better measure is behaviour. Are employees using AI tools in their actual tasks a month later? Has the time spent on routine work dropped? Are managers seeing better first drafts, faster reporting or fewer errors? Onrec has covered this wider point before, noting that training hours are not enough to prove competence and that employers should look at how learning changes performance.
A simple approach is to pick two or three tasks per team before training starts, time them, and check again after six to eight weeks. It is not a perfect measure, but it gives HR leaders real evidence to bring to the board.
What this means for recruiters
Building AI skills in-house also changes how recruitment teams write roles and assess candidates.
Job adverts can shift from asking for “AI experience” to asking for curiosity and a willingness to learn new tools. That widens the talent pool and reduces the pressure to find rare, expensive profiles.
Interviews can focus on how candidates approach new technology, rather than which tools they already know. A practical task, such as asking a candidate to use an AI tool to improve a short piece of work and explain their choices, often says more than a list of skills on a CV.
Employers who invest in training can also use it as a selling point. Many candidates, especially early in their careers, want to know they will be supported to keep up with change. A clear AI learning programme is a strong answer to the question “how will I grow here?”
The bottom line
The AI skills gap will not be closed by job adverts alone. The supply of experienced candidates is too small, the tools change too quickly, and the most valuable knowledge, how the business actually works, already sits with existing staff.
Employers that pair careful hiring with practical, role-based training will be in a stronger position over the next few years. They will spend less competing for scarce talent, keep their teams current as tools change, and give their people a reason to stay. For HR and recruitment leaders, that makes AI skills less of a hiring problem and more of a development plan.
