Mastercard’s annual Cyber and Risk Summit brings together fintech thought leaders to discuss cybersecurity and risk, the latest innovations and future developments. This year’s summit lived up to expectations.
The session AI as a catalyst for transformation provided insights on how AI has transformed Mastercard’s payments ecosystem, its business practices and how it will continue to provide AI solutions for Mastercard’s customers. Topics such as ethical AI, blockchain security and where we stand in the journey toward full automation were discussed. Sudhir Jha, Mastercard SVP and head of Brighterion, and Laura Quevedo, EVP of Transaction and Decision Services at Mastercard, were joined by Ray Wang, Founder, Chairman and Principal Analyst of Constellation Research.
Insight 01: How AI has created safety, security and enhanced customer satisfaction
Coupled with massive data growth resulting from digitization, the past year has brought an even more rapid migration to online services, both personal and e-commerce. This means more connected devices and financial institutions (FIs) are recognizing the need to drive better user experiences for their customers.
How AI transformed Mastercard: safe, secure and enterprise-wide
Mastercard has leveraged the power of AI for over a decade, protecting their customers from fraud without compromising the digital user experience. With the capacity to make 1,000 decisions per second, AI has become the key technology at the foundation of almost everything Mastercard does.
“You always starve for data, for processes that you can actually use AI for, and we have a wealth of that at Mastercard,” said Jha. “We have enabled many different use cases beyond fraud within Mastercard itself.” From network monitoring and cyber risk assessment, to helping customers open new accounts, evaluating credit risk and protecting the transaction ecosystem with the multi-layered connected intelligence strategy, Mastercard is incorporating AI solutions. AI supports Mastercard’s internal budget forecasting, optimizes services, simplifies routing and manages support.
“With AI at the core, we can even extend this capability to cover wider transaction sets across both card networks as well as bank account networks, across open banking, really embracing the whole concept of the Internet of Things. The result is a safer, seamless consumer experience… while ensuring the customer is at the center of what we’re doing,” said Quevedo.
How AI can transform FIs with security and customer-centric use cases
Secure payments and the broader scope of transaction integrity are core values for Mastercard. As a result, Brighterion’s roadmap includes developing use cases that extend AI’s security and customer service benefits to FIs. Customers can leverage Mastercard’s real world AI experience with solutions for credit risk, acquirer fraud, AML and compliance, and healthcare fraud, waste and abuse (FWA). Brighterion will continue to develop AI solutions for customer problems, supporting them as they transform their businesses.
How AI will make blockchain more secure
An exciting revelation by Jha is the prospect of introducing AI to blockchain for both data governance and security. “AI can make cryptocurrency much safer. At Brighterion, we always look at safety, security and accountability,” he said. “New systems have less visibility in terms of who is transacting. Some of these transactions are going to be used for money laundering.”
Insight 02: AI is improving lives through financial inclusion
Mastercard’s mission is to reshape the digital economy so everyone—individuals, financial institutions, governments and businesses—can realize their ambitions.
Using AI for credit risk at origination enables financial inclusion by extending credit to those who may have been denied previously. “We want to broaden the universe where more people get involved in financial transactions and can access capital” to start or grow a business, Jha said. “We want to help communities find ways to increase economic independence.”
AI boosts origination, giving more people access to the system by analyzing their entire financial histories and helping to manage their credit portfolios, predict delinquency and identify when they need help before delinquency occurs.
Insight 03: Ethical AI is transparent and trustworthy
One of the key challenges in taking AI across financial use cases is a high level of skepticism and mistrust. Quevedo stressed that Mastercard is working hard to overcome the challenge by focusing on innovation, inclusion and the ethical use of AI.
Mastercard works to proactively demonstrate the responsible use of technology by implementing governance processes and ensuring compliance, such as minimizing bias in algorithms and making sure they are understandable. “At Mastercard, the ethical use of AI means that it’s trustworthy, it’s transparent, and it’s explainable,” she said.
Insight 04: Privacy—collecting just the right amount of data and sharing only intelligence
With the high volume of data available, it’s become a valuable asset. There’s a race to collect and leverage data and consumers are rightfully concerned by how it’s used and shared. Jha advocated collecting “just the right amount of data.” He gave the example of weather prediction. Why would a weather app require your address when knowing the zip code you are located in is enough? Why does a lender need to know your exact birth date when an age range will suffice? And how much of that is shared, when aggregated data will accomplish the same goal?
Jha said the key is to collect data and derive intelligence from that data. The intelligence is what should be used or shared with data partners, not the data itself, protecting the entirety of the customer’s data and personal privacy. In some instances, such as credit origination, that intelligence may be “combined with other intelligence to create a better application,” he said.
Insight 05: Where we’re at with the five steps of automation
To finish out the session, Wang asked Quevedo and Jha where they think the fintech industry is at in terms of the five steps of automation.
Step 1: Basic automation
Wang compared basic automation to cars with cruise control. The gas starts going on its own, it can hold the speed, but humans still have to drive. “Are we there yet?” he asked.
Jha and Quevedo agreed that, yes, basic automation is being accomplished.
Step 2: Human directed automation
Staying with the motor vehicle analogy, it’s steering on its own, but humans can intervene.
Quevedo believes automation has that ability today, but “how many people let go of the wheel remains to be seen.”
Step 3: Machine intervention
Machine intervention is akin to automatic emergency braking.
“What does it take to get us to that point where—Laura, you’re comfortable in transaction services, Sudhir, the technology’s ready to go?” Wang asked.
That’s definitely where Mastercard wants to go with the latest AI technology capabilities, Quevedo affirmed. “It’s going to take a little bit of time before we’re feeling comfortable enough to allow some of that decisioning to happen. But we’re making progress.”
Jha said Mastercard is making waves. Currently, the AI applications where Mastercard has labeled data are in the fraud applications, and some transactions must be sent for review. “The machine is doing the work, but there is human intervention to make sure everything is going in the right direction and that there are no obvious biases or incorrect decisions made,” he said.
“But not every use case is there. It all depends on how comfortable we are giving that control over, even for a little bit, completely to the machine.”
Steps 4 and 5: Fully autonomous and humans are optional
For the final two stages, “fully autonomous” and “humans are optional,” the group agreed that technology is not ready for them.
“My view is that it’s going to take time. We’ve got to train this, we’ve got to make sure the systems are accurate,” Wang summarized. “Ninety percent accuracy may be great, but it’s the last ten percent that takes time for us to learn, to make sure that everybody is happy with what’s going on.”
What the future holds for AI in fintech
Innovation is only stopped by imagination. For the head of Brighterion, that means using AI as a catalyst for change is a never-ending journey.
“I’m really excited about AI, IoT, blockchain, quantum computing: all of these together and how AI can accelerate their journey but also can leverage from them and accelerate its own journey,” said Jha. “Quantum computing can really enable AI to do things that are not possible today. Similarly, IoT can provide data that is not possible to get today, therefore creating possibilities to do more things in real time and provide real-time insights. It’s the intersection of different technologies.”
And now the very process of building AI models has been transformed by using AI for that very task. With AI Express, Mastercard’s AI model-building process, customers experience the power of AI in as little as two months. The AI Express team can show the customer a quantifiable uplift on results and return on investment (ROI) before going into full production, a previously unattainable metric. Most customers see at least three times improvement in their results.
The AI as a catalyst for transformation session is now available on-demand. Register to learn more from international thought leaders about trust, cybersecurity, securing the payment ecosystem and the latest fraud trends.