Why businesses cannot hire their way out of the AI skills gap

Businesses tend to approach an AI skills shortage as a recruitment problem by buying new tools and hiring a few specialists. As soon as the wider workforce receives access, the business expects productivity to magically improve. But the sparks don’t fly, and the immediate conclusion is that the organisation needs even more scarce expertise.

 

Too often, organisations give employees access to AI and expect them to work out the rest. Specialists can set up the technology and establish policies, but a small expert team cannot teach every employee how to apply AI to their role. This capability has to be developed across the workforce.

 

This results in a significant workplace divide between employees who are taught to use AI effectively and colleagues who are left to work it out for themselves. So, while businesses are quick to blame AI for creating a talent shortage, it is highly likely that they are inadvertently creating the gap.

 

Access does not create equal gains

According to a study of 5,179 customer-support agents, access to a Generative AI assistant can, on average, increase productivity by 14%, with overall improvements in customer sentiment and employee retention. Among novice and lower-skilled workers, the gain reached 34%.

 

It would be easy to think that simply providing access to tools is good enough to increase productivity, yet the result still depends on whether the tool suits the task and whether the employee can understand and apply what it produces.

 

The same research also challenges the assumption that the most experienced or technically confident workers will always gain the most. On the contrary, employers may find considerable untapped capability among people who have had fewer opportunities to build expertise. Structured training can help them use AI as a learning aid within their role. Leaving adoption to personal initiative gives the advantage to employees who already have the confidence or time to experiment.

 

That creates an uneven organisation. A small group becomes increasingly capable, while others may limit their use to drafting generic text or accepting the first answer a system produces. This shows that applying judgment is a skill that can’t be transferred easily.

 

Why external hiring has limits

There will be cases where organisations need to recruit specialist AI expertise. While buying capability can speed up parts of an implementation, it has limited reach. Specialists cannot stand beside every manager to assess a recommendation or review a response.

 

Relying too heavily on recruitment also exposes employers to an expensive labour market. PwC’s AI Jobs Barometer found that jobs requiring AI skills attract wage premiums, while sectors more exposed to AI have recorded stronger labour-productivity growth. So, organisations that leave internal capability underdeveloped will compete externally for scarce skills, while much of their existing workforce will remain unable to realise the value of the tools already purchased.

 

History is already repeating itself – the use of spreadsheets, for example, became widely available, yet ownership of the software did not make every manager equally good at modelling scenarios. Only those who used spreadsheets to the max of their abilities gained true advantage. AI is likely to follow the same pattern – if organisations change how work is organised and prepare employees properly, they will see better results.

 

Teach the role rather than “AI skills”

Generic AI training often begins with employees being shown how to enter a prompt and generate an answer. That may build familiarity, but familiarity is a low bar for competent use. Instead, organisations need to start training with each role. For instance, a manager may need to use AI to summarise information and compare options, while a sales professional may use it to draft faster while preserving customer context and legal compliance. A compliance officer will require a different standard of evidence from either of them.

 

Whatever their role, employees need to understand how to use AI, where it provides the most benefits and which outputs they can trust. Then, they need to apply it to their work over and over again until it becomes second nature. Training needs to also address mistakes that can cause harm, such as entering confidential information into public tools or relying on AI outputs without checking the original sources.

 

Employers also face growing governance expectations. Article 4 of the EU AI Act requires providers and any business using AI in their operation to ensure that staff have an appropriate level of AI literacy, based on their existing knowledge. Therefore, developing employees’ skills can improve performance while reducing the likelihood of unsafe use.

 

Training designed within day-to-day work

A one-off course will struggle to keep pace with tools that are changing so rapidly, as we know. Employers need employees to practise on real tasks, with continuous support available, especially as models develop rapidly and each organisation finds better applications.

 

It’s also important to focus on how employers measure productivity and ROI. They need to examine whether people are using AI in their work and whether output quality improves. Changes in cycle time and rework can show whether a use case is genuinely helping, but it will be possible to see where more support is needed if there are more risk incidents or lower employee confidence.

 

Nearly three-quarters of organisations expected generative AI to change their talent strategies, particularly work processes and upskilling or reskilling. However, employees are unlikely to develop strong habits if AI remains an optional experiment with no connection to how work is managed. Equally, the response cannot depend on recruiting enough specialists to cover gaps. The scale is too large, and the capability required is too closely tied to each person’s role.

 

Businesses already employ people who understand their customers and the decisions behind their work. AI training needs to build on that knowledge. This allows organisations to develop capability across the workforce and reduce their dependence on a small external talent pool. Otherwise, they risk finding out the hard way that the apparent skills shortage was partly a failure to develop the people they already had.

 

---

 

Rodrigo Bolaños, Global AI Strategy and AI Innovation Lab at Mindtools Kineo

 

Main image courtesy of iStockPhoto.com nd saifulasmee chede