To do that, Reiskind has a traditional data management and data governance function, along with a relatively traditional data quality function. In addition, he also oversees AI governance, which isn't always handled by the CDO's office.

However, since Reiskind leverages AI as a tool for innovation, his role is the most suitable within the organization to ensure the responsible development of AI. He then mentions having a data strategy group under his wing that revolves around product development.

Reiskind’s team sits at the forefront of the data organization along with the business and product development teams to foster responsible innovation. He shares that by understanding the business needs from data, the team supports as subject matter experts while maintaining transparency and responsibility.

Moving forward, Reiskind acknowledges that he feels fortunate to be in a company that is one of the early adopters of AI governance. Then, the introduction of data responsibility principles emphasized individual ownership of data and becoming responsible caretakers, he adds.

From the beginning, the organizational leadership had set a clear tone for ensuring the ethical use of data, which has been recently expanded to include the responsible use of AI, says Reiskind. As an organization, he believes that the brand value depends on trust. Trust is at the core of everything at Mastercard, and it is a business value, says Reiskind.

When it comes to key stakeholders, he affirms that his first stakeholders are the businesses that are working with him to develop new products that depend on data. Building on that, Reiskind says that all traditional payment products are data passes – the data passes from a merchant to their bank and then to Mastercard.

Therefore, while people might think of Mastercard as a credit card company, it is essentially a data business at its core. Because of this, the product teams are aware that they need to work with high-quality data to make their products work and satisfy customers.

According to Reiskind, the other key stakeholders are technology teams, and they must understand which data assets should be collected, how to transform data, and the requirements for data access, segregation, and localization. He notes that it is a collaborative partnership to guarantee proper product functioning.

Similarly, the legal and privacy teams are crucial stakeholders for Reiskind, specifically with the expanding privacy and AI regulations. He states that while they provide the requirements for developing responsible AI, they also need to know what the organization is capable of.

None of these are one-way conversations; they’re all part of an ongoing dialogue, says Reiskind. He says that teamwork is highly valued across Mastercard and is exercised regularly as stakeholders come together.

When asked about incorporating GenAI and unstructured data into the framework, Reiskind first affirms working with semi-structured data, such as addresses in transaction data. Due to this, being responsible with data is in the organizational DNA.

Referring to the uncertainty a cardholder faces during transactions, Reiskind states that Mastercard is working to build trust to ensure that cardholders clearly recognize where they spent the money.

When it comes to incorporating GenAI, he affirms tapping into other sources of unstructured data and refers to what he calls “curated data” as a major trustworthy source. Explaining further, he shares an example of making the user manuals and product onboarding applications user-friendly.

By turning these documents into vector databases, Mastercard is creating a safer, curated data asset that is being utilized through chatbots to enhance customer accessibility.

However, while dealing with less structured data, such as providing customer insights from PowerPoint data, internal Word documents, or memos, the team treads with caution, says Reiskind. Currently, someone reviews the output before it is released, but as the company scales the use of GenAI, it is seeking tools that will improve the quality of outputs for these less curated data sets.

Sharing Mastercard’s success story with curated data, Reiskind mentions launching a product called “Shopping Muse.” He explains how this tool uses a retail store’s product catalog, in text or visual form, and converts it into either format for easier searchability on the retailer’s website. It relies on GenAI to handle these text-to-image transitions.

In conclusion, Reiskind states that Mastercard has started utilizing curated data sets for the safe and responsible use of AI. Nevertheless, as technology evolves and use cases grow, the company will get comfortable with uncurated data assets.

CDO Magazine appreciates Andrew Reiskind for sharing his insights with our global community.

PODCAST | Explainable Data Quality Should Be a Thing Just Like Explainable AI — Genesco, VP and CDO