Providing insight into a “first of a kind” AI platform that these financial players have been working on, the speakers – led by Penelope Crosman from Arizent – made a case for global transaction data being leveraged at scale, while at the same time, preserving the integrity of that data and the privacy of cross-border payments.

While it has been evidenced that transformative AI and in particular, machine learning (ML) solutions, can create material business value for banks and in turn, their customers, C3.AI’s Thomas Siebel highlighted that historically, individual financial institutions have been challenged by limitations in data scope.

In addition to this, it has not been fully understood that sharing data can help banks realise the full potential of AI or ML, significantly enhancing the effectiveness and efficiency of shared services like payment screening and fraud detection.

Moving from rule-based methods to data-driven approaches is critical. As Siebel explained, in “this new generation of information technology, elastic cloud computing, the Internet of Things, big data and predictive analytics [is required] to deploy very large-scale enterprise AI applications.”

Siebel also believes that the application of AI in banking and across all industries will establish a “replacement market for everything we’ve done enterprise application software for in the last three decades. This is estimated to be a $30 trillion software market in about four years – the largest market we’ve seen – and I think we have just begun to scratch the surface of the nature of the opportunity.”

Although details of the AI platform were not shared, Siebel went on to say that current use cases of AI include CRM modernisation, revenue forecast creation, customer churn prediction, but associated processes are “erroneous and replete with human error.”

Aggregating data, however, can ensure that the utilisation of this data is accurate, and decisions are based on predictive systems. Further, aggregating non-personally identifiable information (non-PII) data between banks in a secure, encrypted manner can ensure clear forecasts to create the “next best product” and then, the “next best offer,” which is what the AI platform will aim to do.

Reiterating that security is of paramount importance, he added that the financial services industry makes “the US Federal Government defence and intelligence agencies look like amateurs.”

Deutsche Bank’s Rafael Otero opened up about anomaly detection in cross border payments, pattern recognition for anti-money laundering, document flow digitisation and internal uses of AI, ML and natural language processing (NLP). Honing in on the importance of collaboration, Otero said: “We need to have partners like our friends at Swift, that help us challenge the status quo and embark on more journeys [cautiously].”

Following this, Société Générale’s Matthieu Vacarie echoed this sentiment and stated that the French bank currently has approximately 130 AI use cases, with AI usage for fraud detection in payments being successful for at least five years.

Providing a different example, Vacarie explored chatbots and yet, also mentioned that for the application of AI to be successful in this area, “data privacy is a key concern for banks and one of the conditions for us to use AI. We must make sure we protect the privacy and the anonymity of the data that we use and especially in cases where – which we hope to be able to do in future – share among banks.”

In a VT during the session, Tom Zschach, chief innovation officer at Swift, explored this new AI platform. “We see the use of AI and predictive analytics as a key enabler to deliver our strategy, which includes instant and frictionless payments, and gives us further enhancements to better serve our customers and improve the customer experience. We are focusing on building on AI and ML solutions that will enhance our existing products and deliver unique insights and operational efficiencies to the entire network. In all of this, collaboration is key.”

Zschach continued: “We have started develop and validate in network anomaly detection with foundational banking customers and strategic vendors to demonstrate the increased value and operational efficiencies that AI solutions can bring to the marketplace into our products and services together, and combined with explainability of the models. This creates a foundation for growth in AI and ML services and allows us to respond to customer needs in a much more agile way.

“Internally, we're rolling out an enterprise scale AI platform at Swift, with our technology partners. Red Hat is being used for the container platform. Kove, a small, specialised software company has given us software defined memory, and of course C3.AI, we're using for dataops, for machine learning ops and for DevOps.

“This is going to create a powerful, first of a kind, AI platform at Swift and our platform and architecture will scale to the requirements of the ecosystem. It'll be flexible to meet the dynamic needs that AI solutions will require. And the way that we're building it, it will actually make data scientists a lot more productive as well.”