Fireside Chat, Interview
How AI is Powering Payments, with Greg Ulrich of Mastercard
- Greg Ulrich serves as MasterCard's Chief AI and Data Officer, appointed in Q2 of the current year following a career path spanning the nonprofit sector, applied predictive analytics at APT (acquired by MasterCard a decade ago), and leading corporate development/M&A strategy.
- MasterCard distinguishes between traditional AI (machine learning) and Generative AI (Gen AI) based on use cases: traditional AI is preferred for structured data tasks like forecasting and fraud detection due to efficiency and cost, while Gen AI is deployed for unstructured data, knowledge synthesis, and content creation.
- The company's AI strategy is organized into four strategic buckets: "Safer" (ecosystem security), "Smarter" (transaction efficiency and insights), "More Personal" (B2B2C partner personalization), and "Stronger" (internal operational productivity).
- Safer (Fraud & Security):
- Launched "Decision Intelligence," a fraud solution using Gen AI and recurring neural networks to add merchant vector database features to existing models.
- The system analyzes ecosystem-wide merchant behaviors to score transactions, including those with merchants a user has never visited before.
- More Personal (Customer Experience):
- Deployed "Shopping Muse," an AI-powered chatbot that enables in-store style product recommendations for online users via natural language queries.
- Stronger (Internal Operations):
- Implemented a Gen AI digital assistant for customer onboarding that utilizes Retrieval Augmented Generation (RAG) to automate manual tasks and answer technical questions for banks and merchants.
- The onboarding assistant incorporates a "human in the loop" mechanism where chatbot outputs are verified by human agents to ensure accuracy.
- Governance and Data Strategy:
- MasterCard prioritizes data safeguarding and trust as a core competency, refusing to compromise security assets to prove Gen AI value to external partners.
- The company employs a "hub and spoke" organizational model where the central AI team identifies cross-functional opportunities while business units (fraud, HR, finance) lead product ideation and development.
- New AI initiatives are tracked via specific KPIs established at launch to measure ROI, efficiency gains, and customer satisfaction over time.
- Partnering Criteria:
- MasterCard prioritizes early-stage fintech partnerships through its "StartPath" program if the technology fits the "safer, smarter, personal, stronger" framework and adheres to strict governance and security rubrics.
- Decisions on external vendors are made based on feasibility, viability, and the specific problem being solved (e.g., handling unstructured data or manual tasks).
- Market Sentiment and Risks:
- Industry adoption is cautious rather than unbridled, with significant concern regarding AI hallucinations, accuracy, and efficacy in regulated financial environments.
- The company is currently holding back on direct consumer-facing applications of Gen AI, preferring to deploy solutions with human oversight to mitigate risk.
- Future Outlook and Excitement:
- Ulrich identifies multi-modality (integrating text, images, voice, and video to create a single source of truth for tasks like invoice reconciliation) as a key near-term development.
- Evolution in "reasoning models" is cited as a critical trend, specifically the ability of models to confidently state "I don't know" rather than providing confident but incorrect answers.
- Increased focus on transparency, responsibility, and trust is expected to be a dominant trend in the AI sector over the next few years.
- The company views the scaling of models and the distinct value of proprietary data as a primary competitive differentiator for future inference capabilities.