In the rapidly evolving world of financial technology, payment processors are seeking new ways to leverage the expansive datasets at their disposal. Companies like Mastercard (NYSE:MA) and Plaid have recognized the untapped potential within the structured transaction data they possess. These efforts aim not only at enhancing operational efficiency but also at creating robust tools for various applications within commerce and finance. As AI technologies continue to advance, the intersection of financial data and artificial intelligence presents opportunities to redefine industry standards, offering insights that were previously out of reach.
In recent years, the financial sector has witnessed increased adoption of AI models aimed at interpreting the vast volumes of transactional data. Unlike unstructured data like text or images previously employed in AI training, financial records provide structured datasets closely linked to economic performance. With this, in March, Mastercard unveiled its generative AI foundation model. Built on billions of anonymized transactions, it promises to be a potent tool for applications like cybersecurity, personalization, and portfolio optimization. Two weeks later, Plaid introduced its own transaction foundation model, promoting what it defines as “intelligent finance”.
What is Driving This AI Adoption?
For these companies, the main objective is to utilize AI to streamline and enhance tasks across the financial sector. Mastercard highlights the power of their large tabular model, focusing on structured datasets to cover a range of processes from fraud detection to loyalty programs. On the other hand, Plaid concentrates on resolving inconsistencies in transaction data through its sophisticated model. This differentiation aims to provide unique solutions to diverse financial challenges.
Challenges and Advantages of Unique Data
Financial data processing giants like Mastercard process billions of transactions annually, while Plaid connects numerous U.S. institutions. This extensive data access positions them uniquely compared to other AI tech developers. Mastercard’s statement regarding their model’s capability to reduce the complexity of maintaining multiple AI systems stands out.
“The model could reduce the need to build, train and maintain thousands of separate AI models for different markets, use cases and customers,”
Mastercard stated. Meanwhile, Plaid’s focus is on minimizing ambiguities in merchant data, a step that aids in efficient merchant recognition and categorization.
According to Mastercard, its newly developed model is already showcasing improved performance when compared to existing machine learning techniques. By harnessing limited human intervention, this AI can better identify uncommon transactions, which are known to trigger false positives. Plaid, comparing its recent performance, noted significant improvements in income classification and loan payment detection, claiming substantial benefits over previous models.
Despite progress, financial leaders are cautious about the extent of AI deployment. While most surveyed CFOs are open to integrating AI in defined functions, there remains hesitation in broader applications due to data integration and trust concerns. A small percentage of executives are ready to let AI drive recommendations or automate complex financial processes.
Both Mastercard and Plaid’s evolutions in AI demonstrate their commitment to leveraging proprietary data for enhanced financial insights.
“Existing cybersecurity AI models rely on data scientists; our model could learn independently,”
a Mastercard representative mentioned. However, challenges in broad adoption remain, highlighting a potential gap between the potential of these AI models and real-world business applications.
