We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 – 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
According to JoAnn Stonier, chief data officer at Mastercard, the last thing organizations want to do is invest heavily in developing artificial intelligence (AI) tools for use across the enterprise, only to discover that that their insights are “lopsided” – driven by datasets that are incomplete, inaccurate, inconsistent or riddled with bias.
“You need to have a strong, ethical approach to your use of data, a strong commitment to really truthful and accurate and good quality of information, so you don’t have the wrong information in your products and solutions,” she told VentureBeat. If you do that, she explained, “you’re going to be in lockstep with what regulators want. You’re going to be working in the same interest as your customers and their consumers and individuals.”
That sense of responsibility around data guides Stonier, who will be part of the Women in Data & AI breakfast panel at VentureBeat’s in-person executive summit at Transform on July 19 in San Francisco, along with Ya Xu, VP of engineering and head of data at LinkedIn, and Molly Parr, VP, product, digital customer experiences, enterprise products and platforms at Capital One. Sponsored by Capital One, the panel will focus on how the increase of women in data and AI-centric fields has helped shine a light on unconscious bias, and can lead to greater understanding of how to avoid it.
A skill set for data privacy and bias issues
Stonier and her team help Mastercard develop data strategy and work on data governance, data quality and data compliance. Her organization also enables artificial intelligence and machine learning and helps design and operate some of the company’s enterprise data platforms.
Stonier’s unique background prepared her for the broad role of chief data officer at a global leader like Mastercard. Trained as a lawyer, she was also previously an auditor and an accountant and, early on, earned a degree in computer science.
“I didn’t know when I went back to law school that all my skills were going to be tied up into a very neat package as the data world emerged,” she said. “One of the CEOs that I worked for at American Express came to me and said, there’s this new area of law that I think is going to be perfect for you, called privacy. So privacy and data really brought all my legal skills together.”
The power of data principles for AI innovation
A few years ago, she said, MasterCard developed a set of data responsibility principles around privacy and security to guide innovation. “With the rise of AI, we are adding a principle around making sure that we use inclusive datasets, have inclusive employees, develop our algorithms and look at the outcomes that we’re developing,” she said. “We want to make sure we don’t have any flat sides in the data that we use or in the methodologies that we’re using to create our algorithms, and that our outcomes are fair for whatever problem we’re trying to solve.”
AI and machine learning (ML) are key tools for Mastercard, including through the use of algorithms meant to spot fraud. “We have dedicated ourselves to constantly refining our information so we can spot fraud at a really specific level and also not have declines that are really legitimate transactions,” Stonier explained.
Andrew Ng, she added, talks about data as the “food for AI.” “So are our datasets the right ones to use – are they representative of the populations that we’re trying to serve?” she said. “Then, what about the algorithms we’re creating, have they been structured, is the algorithmic inquiry the right one? And do they have any inherent biases in just how we’re structuring them?”
Stonier explained that Mastercard has a review process that includes the data design team, data quality team and data science teams that work together in an iterative process as AI and ML methodologies are developed. As they are applied, they are done in a test environment first before being released into production.
“Then we look at the outcomes – we look at the models for drift and anything else that’s going to show us that we’re not getting the results that make sense,” she said. “If we are getting results that show that a certain population is being impacted and it doesn’t make sense, we’ll go back to the drawing board.”
Mastercard’s next-generation network
Stonier also works closely with Ed McLaughlin, president of operations and technology at Mastercard. The two are working to change the way the company’s infrastructure works, including a project they are loosely calling a next-generation network.
“It’s very data-driven in how we’re thinking about the information we will need to process around the world, so we’re working closely together to think through the design, as we know Mastercard’s data needs will change in the future now that we are a multirail type of network,” she said.
The company will also have different regulations to be mindful of going forward. “There will also be different ways that we believe people are going to interact in the new virtual world – with virtual reality and augmented reality,” she said. “We’re going to have to capture all of that information in different ways.”
Stonier emphasized that she does not navigate her career thinking of being female first. “I just think of myself as a business executive, and as a data professional and now certainly as an AI professional,” she said. Yet, she admits there is still bias – and way too many meetings where she is the only female in the room.
“That does have to change, and for AI in particular, we need diverse perspectives,” she said. “Bias can creep in and diverse perspectives are really important as we design machine learning and artificial intelligence for the next generations.”
Events like Transform’s Women in AI breakfast are important, she added, because “The last thing we all want, and I think this is true for men as well as women, is to develop artificial intelligence that learns from itself and leaves the female perspective, or any perspective, out,” she said. “We want our machine learning to actually reflect all aspects of society – so I think a breakfast like this helps remind us that we have to develop AI with that kind of 360-degree lens that really understands all of the different aspects of society.”