There is hardly a business in the region that is not at least considering how to integrate artificial intelligence (AI) into its technology mix. But two main stumbling blocks present themselves — a talent shortage and the identification of use cases. In the United Arab Emirates (UAE) an entry-level data scientist can expect to earn around AED12,000 a month, and salaries can reach up to AED30,000 for a senior data scientist and more than AED 50,000 for a lead data scientist. These are rare — and hence, expensive — skills. The top salaries are going to people who have both the technical know-how and business knowledge to overcome both stumbling blocks. I need hardly tell you how rare these professionals are.
Rather than go on a unicorn hunt, many enterprises are choosing to upskill the talent they have, training AI experts in business matters and introducing business users to the world of AI. A recent Dataiku study revealed that among businesses that have implemented AI at scale (delivery of what we call “Everyday AI”), 85% had cross-trained their teams. But their success was only made possible by ensuring business users got the same access to data and AI as technical teams did. Other recent Dataiku research found that while almost all (94%) data science and technology specialists in the UAE have this access, only 42% of business users reported the same.
The solution to this gap lies in how organizations start their AI journeys. They commonly find that they operate a series of information silos, which must be eliminated before the real work can begin. Part of this process may involve developing a shared infrastructure designed for reducing costs and time to value. An AI Center of Excellence (CoE), composed of varied and complementary talents will work towards building a team of unicorns. They will develop AI products, stay current on technology changes, and nurture AI champions across business units. If the CoE demonstrates value, wider programs may emerge, and AI may be adopted and used more widely and for more critical tasks.
Read: Sheikh Hamdan bin Mohammed announces opening of the Dubai Centre for Artificial Intelligence
Hub and spoke
At this stage, silos will irrevocably break down and cultural challenges may emerge but with data and AI skills on the rise along with value, the bulk of concerns should erode to make way for a sharing environment where value can be added everywhere. What is created in place of legacy silos is a hub-and-spoke super-team with AI experts in the center and business units all around. The old request-and-delivery working model is replaced with collaboration between hub and spokes on every project, the ownership of which is retained by the spokes.
To ramp up AI adoption, the CoE is supported by a Center for Acceleration, which is responsible for getting frontline business users involved in product development. The goal is to build unicorn teams in every spoke so that subject matter experts can bring their knowledge to the program and innovate without having to file a request with the IT and AI teams and wait for a visit from a requirements analyst. Business users as AI developers can add value quickly and become pivotal in introducing flexibility and rapid ROI across the enterprise.
When democratizing access to data and AI, an embedded structure (where rules such as responsible AI are simplified and centralized and data science is integral to every business function) works best. This is easier for companies that started as digital businesses, but for legacy organizations, it is a process. No matter which category applies, a common AI platform makes life considerably easier. From data collection to experimentation, training, analysis, and development, the platform will facilitate the building of unicorn teams and the interdisciplinary working approach that follows.
Experiment and learn
Low- and no-code development platforms not only make AI more accessible to business users; they help companies to align with programs such as the UAE’s National Program for Coders. Self-sufficiency emerges from a talent crisis and ensures that progress is not hindered, either within a company or in the wider economy. The platform must entice users of all skill levels to experiment and learn. If it can do this, the CoE will show the value of AI quickly and clearly and embed it in the DNA of the business. It is the central platform and the dedication of the CoE that will turn adoption into longevity. At the end of this road is Everyday AI — the culture where every employee uses AI in much the same way as they use email.
Another detracting argument may arise — that upskilling employees only means they will leave and take those skills with them. But in today’s “employee experience” labor market, what we often see is polls suggesting workers will leave to find an employer that will invest in their professional development. This means that not investing in people is a greater risk than investing. Even if untrained employees stayed, what value would they add to a company’s struggle to stay relevant in a digital economy?
A common AI platform can play a central role in upskilling, either through training groups of staff with the same skills from different departments, or training differently skilled employees from the same department. The central AI platform should inspire adoption and facilitate upskilling, however it is handled. This is how an enterprise can build its organization-wide AI dream team of unicorns.
A December 2021 McKinsey global study on general business transformation showed that among change projects where frontline employees felt “a sense of ownership” and took “the initiative to drive change”, 79% were successful. As the democratization of access to data and AI increases, so does upskilling. And as upskilling proceeds, the awareness of the value of AI increases. This calls for more upskilling. And the cycle repeats to deliver a range of benefits across the business — a sustainable culture of Everyday AI.
Gregory Herbert is SVP & GM – EMEA at Dataiku
For more tech news, click here.
