Fraud fighting AI

Using AI to fight fraud: a Conversation with MindBridge CTO Robin Grosset

In Featured, Insights by Carta Worldwide

MindbridgeThe high-technology CTO talks about using AI to scale expertise and fight fraud.

Robin Grosset has an impressive background, with degrees in Physics and Computer Science, and a résumé that includes Chief Architect at IBM Watson Analytics. Now, he’s helping to build the world’s first Artificial Intelligence Auditor with MindBridge. We asked him about using AI to get richer insights, and how auditors will interact with the Internet of Things.

To learn more, join Robin and other fintech leaders at Mobey Day in Toronto, August 30–31, 2017.

Carta: What are the most important applications for Big Data and AI in finance right now?

Robin: Helping people keep up with an impossible amount of data. For example, the accounting/auditing profession has been around for centuries. The first auditors started a millennia ago. But the rate of data collected is increasing all the time. Today, a human being couldn’t possibly look at all the data. So people developed an approach called “sampling” to look at a small number of transactions. If the sample was okay, they’d reason the rest of the transactions were okay. But if you’re looking for an anomaly, and if you’re looking for fraud, you might not find it in a sample.

If you can capture the expert knowledge an auditor has, you can codify that into a machine, and it can go look at every single transaction for you. If you’re only sampling, you could miss fraud for years. That’s why MindBridge exists. Now, more than ever before, it’s possible for machines to do that auditing, and do it well.

We’re going to see AI step up and take on the role of processing more and more information where a human being could not. It takes a long time to scale expertise. However you can make a machine that has expertise, and only defer the exceptional cases to the human.

Carta: How can AI help financial institutions get beyond number crunching to value-added insight?

Robin: An insight for one person is slightly different for different people. Financial institutions are struggling with how to analyze information. The big bottleneck was always human capital. Data scientists were the scarcest resource, since they’re in short supply. So can we make machines that do analysis, like having a data scientist in the box to help non-technical professionals?

With AI, like never before, you have the facility for AI to perform the analysis, but then you have to understand how the machine communicates. With predictive learning, a machine would say “yes” or “no,” but the human couldn’t extract value.

AI is moving towards “explainable AI” that can communicate on a human level, in quite a bit of detail. Deep learning AI will be able to communicate insights so a human can understand. It’s critical in finance that the human operator be able to not just get the “answer” the AI produced, but how it arrived there. Like in school, you got one point for getting the answer right, and one point for explaining why. I think that’s where the promise lies, actually. It’s in the clear explanation of the answers.

Auditors are trained to be skeptical. And periodically they’re audited by a regulator. They have to be able to stand behind all their decisions, and have a logical reason why they made that decision. So they need that level of explanation and assurance. It can’t be a black box.

Carta: What role will the Internet of Things play in collecting even more data for auditors?

Robin: IoT will exhaust data such as voice and location data that isn’t currently available. For example, if a trader is standing on the street and gets a stock tip that he shouldn’t have, IoT may be able to place those two individuals in the same place, and through cell phone data may even be able to transcribe the conversation. That gives auditors one more piece of data to work with.

In Bangladesh, there was an instance of bank fraud where the fraudsters stopped all the printers from working, which distracted people, so they could more easily cover their tracks. IoT could help connect seemingly disparate systems, such as printers, to trigger the system that there’s an anomaly.

IoT is going to make use of all this data in useful ways, particularly for combating fraud in finance.

Carta: What’s next in terms of AI innovation for financial services?

Robin: There’s already been huge adoption of AI across the industry. But one piece that I would like to see happen is higher-level cognitive functions being fulfilled by AI, where it would be easier and easier for people to do their jobs because they’re being helped by AI functions. I’d like to see people accomplishing more in less time, with a higher degree of accuracy.

Join Robin and other fintech leaders in Toronto at Mobey Day 2017, on August 30–31. He’ll be speaking about conversational AI for banks. Register online.