Quantexa (see Pulse 2.0 profile with CEO Vishal Marria here) is a global data and analytics company pioneering Contextual Decision Intelligence and it empowers organizations to make operational decisions by analyzing meaningful data. Pulse 2.0 interviewed Quantexa Head of Financial Crime Solutions Clark Frogley, a former FBI Supervisory Special Agent, to discuss how financial institutions can utilize artificial intelligence (AI) to tackle fraud in banking.
Amit Chowdhry (Pulse 2.0): With advances in AI having continued to gather pace, what are the key risks and challenges banks are facing as a result?
Clark Frogley (Quantexa): One of the biggest challenges financial institutions face as AI rapidly advances is managing the complexity involved in deploying it with the necessary levels of transparency required to build trust and meet governance requirements. In an industry that’s highly regulated, banks need to take a white box approach to AI, so it’s clear how AI models are developed, trained, and deployed, and they can be certain they’re based on a solid data foundation.
AI models do not work on a one-size-fits-all basis – specificity is important, and an AI model is best designed and deployed to fit a specific application domain. Evaluating where to deploy AI and how best to deploy it to solve that specific problem while ensuring transparency in how the AI model works are key, yet continue to be a challenge for many organizations. According to Gartner, just 54% of AI projects on average, make it from pilot to production.
They also face significant data challenges when leveraging AI. AI models are only as good as the quality of the data they’re based on. Many financial institutions are dealing with massive amounts of data from siloed datasets, often resulting in inconsistency across sources, duplicates, and outdated information. The complexities of dealing with unreliable, siloed, and duplicated data can impact confidence in decision-making. Research by Quantexa suggests that fewer than a quarter of IT decision-makers believe their organization trusts the accuracy of the data available to them. Without a solid foundation of reliable data to feed into their AI models, banks will struggle to drive the accuracy they need from AI. This accuracy is essential for driving Decision Intelligence, connecting data across silos and formats to create a single, trusted, and connected data resource to power accurate AI-powered decision-making at scale.
Amit Chowdhry (Pulse 2.0): How does AI help banks better identify and tackle potential fraudulent activity?
Clark Frogley (Quantexa): Bad actors are leveraging the latest technologies to deploy new methods of committing fraud at higher rates than ever before, and so it is vital that banks continue to remain vigilant. However, many struggle to keep up with the rapid pace of fraud today. As a result, many are turning to AI to develop and roll out new models for identifying patterns of fraudulent behavior much faster. AI can model what normal behavior patterns look like and test outside of those norms to identify outlier behavior that could signify fraud. It can also understand relationships between individuals and businesses to uncover where there may be hidden or obscure connections that could be signs of fraud.
Banks are utilizing AI to gain advantages over bad actors by using data more efficiently and effectively, ingesting massive amounts of data, and building networks using broader contextual information. This improves the accuracy of AI models for identifying patterns of fraud, providing crucial information that helps banks drive more informed decision-making, risk identification, and faster response. When it comes to identifying fraudulent activity, context is everything. Contextual Decision Intelligence enables financial institutions to go from “guess-work” recommendations to informed and confident decisions derived from trusted data.
Amit Chowdhry (Pulse 2.0): How can banks ensure they are identifying truly suspicious activity associated with fraud and money laundering, and not generating false flags around innocent banking customers?
Clark Frogley (Quantexa): Context plays a critical role in driving accurate AI-powered decision-making. Improving the accuracy of fraud identification requires moving away from binary decision-making to an approach that is based on a more three-dimensional view using disparate data sources that are often out of reach for investigation or compliance teams. AI improves banks’ ability to bring additional contextual information from a wider range of data sources into their decision-making. Contextual Decision Intelligence supports more nuanced decision-making to provide confidence in the AI systems that are automating the identification of fraud to enable faster, more accurate decisions. For example, knowing that a large transfer between two bank accounts was made between two people who are related as family members poses a lower risk than other non-family transactions, thus not something to flag as suspicious. For AI to effectively and accurately identify patterns of fraudulent behavior, it requires both dependable data and contextual information combined with the right alerting logic.
Taking a composite AI approach can also help to reduce false flags. If banks are only relying on one AI model or technique to fuel their decision-making, it can lead to limitations in performance. By using composite AI, which blends together different AI techniques – for example, AI, Machine Learning (ML), and Natural language processing (NLP) – together with their domain expertise and contextual data, banks can build more performant analytics models that drive more accurate and intelligent fraud detection, fewer false positives and more accurate true positives.
Amit Chowdhry (Pulse 2.0): Why do you believe it is important that banks invest in new technology tools to combat fraud?
Clark Frogley (Quantexa): As fraud continues to become increasingly sophisticated, with bad actors deploying new approaches and new technologies, it is imperative that banks take an aggressive approach to combating this growing threat. In the ongoing race against fraudulent behavior, banks must respond in the same manner as their evolving adversaries and do so by leveraging the latest technologies at their disposal to keep up. As fraud volumes continue to rise, it’s becoming more and more difficult for banks to resolve through human intervention alone – the same level of technological firepower is required to respond to and prevent these risks.
