In his keynote address at Data Science Day 2025, hosted by the Columbia University Data Science Institute, Rick Rioboli offered a candid look at the realities of deploying AI at scale inside a large, operationally complex company.
It was a talk grounded not in possibility, but in practice: what it actually takes to move from experimentation to adoption, and from promising tools to systems that meaningfully change how work gets done.
While interest in AI has surged across industries, Rioboli noted that few organizations have yet made the leap from pilot projects to broad implementation. In many cases, the models work. What’s missing is alignment—with workflows, platforms, and people.
“If we don’t get enterprise adoption,” he said, “the economics break down.”
For AI to deliver value, it needs more than performance. It needs a place to live.
As Comcast has introduced AI into areas like customer service, network optimization, and software development, Rioboli observed that technical capability alone isn’t enough. People need to understand what a system is doing—and feel supported by it.
“Predictability matters,” he said. “Not just for performance, but for confidence.”
In his view, trust isn’t just an ethical consideration. It’s infrastructure.
Rioboli emphasized that successful AI adoption requires systems that fit the environment they’re entering—not just powerful models, but durable pipelines, quality data, and interfaces that make sense in context.
Large-scale transformation, he suggested, is rarely driven by the flashiest technology. It’s built on structure, coordination, and long-term investment in how tools interact with the people who use them.
This keynote traced a path not from idea to breakthrough, but from insight to implementation. Rioboli didn’t overpromise. Instead, he made a case for the kind of work that often goes unseen: aligning infrastructure, building trust, and giving systems the time and space they need to prove their value.