Why upskilling is the antidote to the ‘AI skills gap’ ailment

Whether they use the term or have even heard the term, everyone wants Everyday AI — the culture where artificial intelligence (AI) is a natural and habitual go-to for resolving business challenges, and where governance is at a high enough standard to all but guarantee value is added by every project. The UAE stands to receive a 14 per cent contribution to 2030 GDP from AI activity.

In a regional business community overflowing with AI use cases, each enterprise has leaders who understand this. And they further understand that to not be a part of the AI subplot may mean being left out of the larger economic success story of which it is a part.

But scaling up toe-in-the-water exploration of AI to widespread, integral adoption in every corporate discipline is not without its challenges. The basic problem is twofold.

One, regional AI skills gaps mean a lack of talent to execute individual projects. And two, new hires may be conversant in AI but may lack the business knowledge to identify use cases. It is because of this dual obstacle that some businesses are abandoning arduous recruitment drives that seek out the unrealistic combo package of AI and business skills in a single candidate, in favour of upskilling inhouse business analysts in AI.

I shall leave it to you to trawl through the many studies that tell us there are way more data-science jobs out there than there are data scientists. If you are currently looking for a data scientist to propel your organisation to new heights, you will have discovered some version of this for yourself. But you cannot wait around for this situation to resolve itself. You have a vision for Everyday AI.

The answer: upskilling

While you may have considered outsourcing or consultancy, neither resolves the domain-knowledge gap. And while you may fear talent attrition in the wake of AI upskilling, the risk of this has been shown to be small.

In fact, there is more evidence to suggest that attrition will result from knowledge workers seeking out opportunities where they can bolster their STEM skills and become more relevant to the digital economy.

So ubiquitous is the consensus that internal upskilling programmes can be the answer to AI skills gaps that modern AI platforms are starting to cater to professional-development paths for all levels of knowledge worker, from the intermediate spreadsheet user to the most knowledgeable data scientist.

Some organisations may opt for the interdisciplinary upskilling programme, where skills rather than department determine who is trained together with whom.

Excel power users from finance would likely find themselves sitting in classrooms with Excel power users from warehousing in this training model. This approach is useful to train large groups at a time and to let a little domain knowledge from each business unit seep into others.

The alternative is functional upskilling, where the use case is the focus, and employees from the same department with similar business skills but differing tech skills — spreadsheet users and database administrators from finance, for example — learn together so they may address a specific problem or set of problems. This approach is used where rapid adoption and time to value are the priorities.

Governance is critical

Whatever approach is applied, stakeholders will soon see Everyday AI taking shape. With a common AI platform, administrators will be able to see projects, models, and data access evolve and will have the means to monitor and oversee all AI-related activity. This is critical for many reasons. Some projects may be found to conflict with the business goals of others, or with the ambitions of the company as a whole, or with the requirements of regulators.

Governance is key to effective talent acquisition and plays an important role in retention rates. It can help prevent costly mistakes as AI users hone their skills, and it can present clear information about data-worker performance to managers who may not be entirely familiar with every tool in use. Both of these capabilities pave the way for a more knowledge-based way of identifying AI talent, thereby increasing trust across the workforce.

Robust governance, delivered by a common AI platform, also helps with comparing business units so that talent may be sifted out of every corner of the company.

Many useful metrics emerge from the common-platform approach. What we measure is, in fact, no different for an AI and ML upskilling programme than it would be for other digitisation projects — ROI, tech budget percentage spent on AI and ML, AI assets produced per data worker per quarter, talent retention rates, and so on.

But it is also important to measure operational aspects such as the percentage of data workers using AI and ML, and the percentage that use a common platform. Retention rates also need to be studied over time, including at the three-, six-, 12-, and 24-month milestones.

Common platform

Each of these metrics has its use in keeping the Everyday AI train on track. For example, we must know the proportion of a workforce that adopts AI and ML because if participation is too low, a sustainable AI culture cannot flourish, and our upskilling investment ends up bearing no fruit and leading to negative ROI.

Common AI platforms are the ideal support element for upskilling programmes because they engender the level of inter-team collaboration necessary to gain momentum in an organisation’s effort to build an AI culture.

The ideal platform will lead to high reuse and automation. It will allow beginners to deliver genuine value quickly while allowing those who like to code to indulge that preference. But the results are the same: working models that add value to the business while empowering their creators and never straying far from core business objectives. In a nutshell: Everyday AI.

Gregory Herbert is the SVP and GM – EMEA at Dataiku

Read: Here’s how corporate upskilling benefits the workforce