Digital transformation and AI walk into a bar…

AI is probably the buzziest buzzword in the payments industry right now. Step aside customer service chat-bots and fraud-based machine learning, AI has a new purpose: driving digital transformation as we enter the fifth industrial revolution.

It is easy to get swept up in the hype, especially with the significant headway that has been made over the last few years into customer management, fraud detection and payments optimisation. Many FIs are dipping an exploratory toe into the AI waters to see what else can be done to increase operational efficiencies and lighten the load. From compliance to code generation, recruitment to orchestration; nothing is off the table. According to a recent report by McKinsey, disruptive AI technologies can dramatically improve an FI’s ability to achieve four key outcomes: higher profits, at-scale personalisation, distinctive omnichannel experiences, and rapid innovation cycles. What’s not to like?

Let’s take a minute to examine the other side of the coin. AI is two things: a tool and a magnifying glass. As we all know, tools are only as good as the people welding them and magnifying glasses amplify and bring to light the smallest details. What does this mean for digital transformation? It means if you truly want to make AI part of your business strategy, you need to get your house in order first.

If we go back to the basics, we all know that your business is only as good as the foundation it’s built on, just as your technology stack is only as good as it’s core, mission-critical systems. Over the last 20+ years, many FIs have added to their tech stacks to quickly transform their businesses and remain at the forefront of the market. Integration layers, plugins, wrappers and all the exciting third-party integrations have enabled these essential quick-fixes but have also created issues for the integrity of the overall architectural infrastructure: spaghetti systems with multiple points of failure and most importantly for AI, data silos.

For AI models to be successful, they must rely the accuracy of the data, the quality of their instructions and the ability to apply algorithms to sensitive use cases to anticipate a balanced and unbiased outcome. Algorithms, and the decisions that are made based on them, ultimately magnify the flaws of their creators. It comes down to data trustworthiness, and if, for example, the source of the data being used is from former data siloes compiled into one large data lake housing both data gold, but also a good portion of human-inputted inaccuracies, replications, and old information, then you are already in trouble.

FIs need to ensure that AI is integrated with modern, future-proof technology that works effectively. The preparedness of the existing technology stack is key and the dangers of adopting AI too early can have significant repercussions. FIs need to review and adapt to new business models with new operations, whilst fintechs need to consider adapting to technologies such as cloud applications, and traditional banks on legacy systems with huge silos of data and poor integration – well, they need to rethink everything.