Mastercard is integrating artificial intelligence into every aspect of its products and services. More than a differentiating factor, it’s about harnessing the power of AI by using it for planning, protecting, and doing good things for others.
On June 17th, the Mastercard Virtual Cyber & Risk Summit was held, wrapping up a series of webinars called the InConversation Series. In these times of working from home and physical distancing, the Summit gave Mastercard the venue to provide education about cybercrime, talk about recent innovations, and network with current and future customers. Michael Miebach, Mastercard president and incoming CEO, welcomed attendees and introduced the opening address by Ajay Bhalla, President, Cybersecurity and Intelligence Solutions, Mastercard.
- The current state of disruption in fraud and cybersecurity
- Businesses under siege: the latest fraud trends & impacts
- Revolutionizing fraud detection: connected intelligence meets CARTA
- Power talk: the value of digital receipts
- Protecting trust across new frontiers: expanding ecosystem security for small businesses
- Connected Intelligence in practice: bridging authorization and authentication
- Future of authentication
- Combating economic crime: uncovering fraud and money laundering
- Cyber futures
Harnessing the power of AI
It’s hard to explain all the ways Mastercard integrates AI into every aspect of its business, but that’s what we managed to do in the session “Harnessing the power of AI.” Sudhir Jha, Mastercard Senior Vice President and Head of Brighterion, and Nitendra Rajput, Vice President, Product Development, Mastercard, gave lots of real life examples. The session was moderated by Karen Webster, CEO, Market Platform Dynamics.
Sudhir Jha set us on the right course by explaining the role of AI in Mastercard’s portfolio.
“Mastercard uses AI as one of the new technology trend initiatives that are fundamentally changing businesses around the world, including Mastercard. We are fully adopting AI into every product and service that we offer, and internally as well,” Sudhir explained. “It’s one of those essential technologies that a company is going to use to advantage, as well as for scaling businesses in every part of the world.”
Sudhir and Nitendra Rajput summed up the overall plan for four areas of focus:
1. AI drives Mastercard’s solutions
Mastercard has been using and investing in artificial intelligence for almost 20 years. For example, Mastercard had been using Brighterion’s AI to detect and prevent fraud and for AML for almost two decades before acquiring Brighterion in 2017. Stronger than ever, Brighterion continues to provide our advanced AI to organizations around the world.
With the support of Mastercard, we can continue to expand our product and offer advanced use cases. Along with Brighterion, Mastercard has acquired other solutions that result in end-to-end protection for consumers, merchants, financial institutions, and governments.
2. AI can be used for custom solutions to unique problems
Sudhir talked about building AI models that can be customized to an organization’s needs, including areas that require more focus during the COVID-19 outbreak. Careful credit risk and delinquency management enables credit managers to address each customer’s needs, preserving valuable client relationships. He also cited using AI to help acquirers manage risky merchants and transactions during what may be challenging financial times for people.
Most important in these functions is the ability to build a model that directly addresses each organization’s needs and helps them prevent fraud, increase success rates, or predict future events based on past data.
3. Harnessing the power of AI to increase productivity
The key to any organization is its people, Nitendra reminded attendees, and integrating AI into human resources practices makes sense. It can scan people’s files to determine those with the right skillsets for particular jobs, determine what skillsets are required for advancement in the company, and find courses to help the person upgrade.
Mastercard uses AI internally for revenue prediction. Using data based on past performance, AI can direct corporate leaders on well-performing units, areas that need more improvement, product development, and return on investment.
4. Using AI for good
Mastercard uses AI to solve big problems. One example is using AI to assign jobs to employees who have volunteered to help non-profits, matching skills and interests with available opportunities.
Nitendra outlined a project where Mastercard used AI to help coronavirus researchers comb through 100,000 articles to ask certain questions, such as the latency of the virus and past fatalities. What could have been months of work when done manually was completed in just a few days.
“It was a fascinating experience for us,” Nitendra reflected.
They were able to predict the number of hospital beds that would be needed, which businesses would survive the pandemic, and more.
How viable are AI models in the face of these challenging times?
Karen Webster posed a question based on what she calls the “elephant in the room.” How viable are these AI models and algorithms when we are seeing so many different behaviors on the part of nearly every single person in the world at the same time? Can existing work still be effective in developing models that predict outcomes?
According to Sudhir, it continues to be about data quality. “In supervised AI models with labeled data, if your data is either biased or incomplete, your model won’t be able to do the right thing,” he said.
“You want to make sure you are using the kind of data you can rely on. So with credit risk, for example, some of the payments data is not very useful because most companies are (during COVID) getting some kind of a holiday for that,” he explained. “You want to make sure that your model isn’t relying on that dataset to provide value. You need to provide more (robust) models that don’t just rely on that data. Unsupervised AI modeling presents less issues. It’s all about the foundation your model is built on.”
Nitendra added, “The payments industry is used to this because in fraud and money laundering, things keep changing all the time. If fraudsters don’t change, it’s easy to catch them.”
Mastercard’s AI engineers have to figure out how to keep adapting and moving along as situations change. “Good practices really help us take care of these dynamic situations,” he said.
Preventing bias in AI model building and showing ROI
AI model bias is always a hot topic. So frequently discussed, in fact, that Sudhir in the past has spoken about it in one of our AI Innovators videos and a previous blog. He says there is very strict governance for model building with checkpoints along the way to minimize or eliminate bias.
That said, Nitendra sees an opportunity. Using AI to build the model with the customer’s own data, taking the time to test it over a six to eight-week period, and identifying the ROI are all part of the development pipeline.
This six to eight-week process, known as AI Express, is fundamental to how the model is built. The team builds and tests the model for the customer’s goals, then works with the customer for a month to help them navigate and use the system. By this time, the model is already showing ROI.
AI Express is a time for innovation, creativity, and evaluation. “If data is the oil that powers the digital economy, artificial intelligence is the refinery,” Ajay Bhalla once said. AI Express is the key to your organization’s AI test drive.