The UAE is home to an almost entirely banked population — 99.9% penetration, according to one estimate from Statista Market Insights and the World Bank. And nearly 70% of us use online banking, a testament to the digital maturity of the nation’s FSI sector. Banks here have access to exponentially deepening and widening oceans of data. The challenge for 2024 is juggling this position with other balls, such as emerging new risks and technologies amid continued market volatility.
Rising to the challenge means using data and AI responsibly (within regulatory boundaries) and transparently (to engender trust). Banks must pursue three broad goals to squeeze optimum returns from their data projects.
Reduce risk
UAE banks are subject to many regulations designed to protect customer data. From the UAE government’s Personal Data Protection Law (Federal Decree-Law No. 45 of 2021) to the Payment Card Industry’s Data Security Standard (PCI DSS), the nation’s financial institutions must be able to clearly “show their work”. They must be transparent and auditable. This applies to machine learning models, so banks must understand how responsible AI relates to governance and MLOps.
This calls for the right skills and processes to be in place, supported by robust AI platforms on governance and MLOps capabilities. As banks’ workforces grow, the complexity of maintaining a mixture of tools will increasingly become a recipe for non-compliance. Project leads must spend time integrating data pipelines if the bank uses one tool for data preparation, another for building models, and another for validation and deployment. Errors will then creep in from all sides at all stages. This will inevitably lead to risks such as missed deadlines, data losses, and security issues.
Maximise ROI
In AI, costs arise in many forms. We have those associated with skills acquisition and retention. We have those associated with implementation. There is the cost of tool procurement. Additionally, we must consider any delays in cost savings because of the wrong systems being in place to allow rapid deployment of a solution. Missed savings are, after all, costly. Organisations must review the entire data and analytics stack to ensure it provides for seamless workflows. Data access and preparation should be integrated into downstream systems so technical teams can take over the work of analysts and easily apply machine learning techniques.
By taking a cost-centric look at the building blocks of the AI stack, teams can identify opportunities for automation of steps in the lifecycle, leading to greater efficiency and reduced costs. Remember that familiar and comfortable tools may be cash sinkholes. Also, at this stage, MLOps teams must think about the cost of model maintenance. It takes commitment to manage model drift and ensure out-of-date artefacts do not harm the business by offering up misleading information based on stale data. Teams must look for ways to be more efficient at maintenance so that costs do not spiral out of control.
ML projects can be dizzying rides filled with the clamour of many stakeholders. To streamline processes, manual work (and rework) should be automated to save time and budget and free innovative humans to work on more high-value tasks.
Empower people
Brands in all industries tell their customers some variant of “we are our people”. In the modern UAE banking industry, relationships with employees are more important than ever. It is critical that institutions retain talent, and one way to do that is to get everyone involved in AI — to build an everyday AI culture. Given the region’s STEM skills shortages, it is risky and expensive to go on the hunt for golden-goose data scientists. There is an opportunity to build an AI team from within, from those who already possess the business knowledge, to quickly identify relevant use cases and add value.
In 2024, we are likely to see financial businesses try to break free from the region’s talent shortages with a grow-your-own approach that concentrates on tooling and employee empowerment. As banks incorporate more AI-based techniques and the centralised, governance-oriented platforms that support them, they will need fewer and fewer data scientists. Instead, business users such as actuaries and quants will emerge. They have a lot to contribute, given their affinity with mathematical techniques, and in turn, AI can significantly augment those roles.
It is at this juncture that Everyday AI comes into its own. By looking at your skills within the context of your business, you begin to see how AI models can supercharge many roles and functions. With the right centralised tools, value is quickly harvested from within rather than waiting for months to hire a costly new data scientist, who will then take many more months to learn the ins and outs of the business. Training AI teams from within leads to quicker identification of relevant use cases, better models, more accurate results, and more practical insights.
Working for you
There is a lot of pressure to juggle market volatility, technology procurement, regulatory growth, and the rest. Data nestles within UAE banks are waiting to be used, but many institutions have not yet implemented the governance and tools necessary to empower their employees. It takes a bold vision to inspire bold steps and make your data work for you. Everyday AI is just a vision that concentrates on risk reduction, ROI maximisation, and employee empowerment so everyone can reap the rewards.
