The Foundations of Scalable AI

MIT Technology Review’s EmTech Future conference is making a simple point: scalable AI doesn’t come from a clever model alone. It needs data, engineering, platforms, operating models, and governance to line up and behave like one system. That is the foundation, not the garnish.

The session is framed around a hard setting: a highly regulated environment. That matters because the easy version of AI deployment is the demo. The difficult version is the one that has to survive compliance checks, internal controls, and the messy reality of real users.

Jing Wang, a principal at Vanguard’s Center for AI and Insights, is the speaker behind the session. Her team works across AI, machine learning, generative AI, and behavioral science, with the goal of turning emerging tools into trusted solutions that improve client experiences, investor outcomes, and business value. Vanguard is clearly betting that responsible implementation is not a brake pedal; it is the way these systems become durable enough to matter.

The session also points to agentic systems, one of the newer ideas getting attention in enterprise AI. But the message here is less about flash and more about plumbing. If the foundation is weak, the ambition won’t hold. If it is strong, then the organization can actually use new capabilities without treating them like a science fair project.