I speak about designing, deploying, evaluating, and governing AI systems in settings where constraints are real, failure is common, and outcomes matter. My talks focus on what works—and what doesn’t. A lot of AI conversations focus on what models can do. In practice, the harder questions are whether anyone will use the system, whether it changes decisions, and whether it improves outcomes. How do you design such systems? How do you know if they work? And how do you govern them?

Talk Topics

  • AI That Works (and AI That Doesn’t): Lessons from Real-World Deployments

  • Why most AI projects fail (and why that’s predictable and avoidable)
    A practical look at why AI systems often don’t translate into impact, and what distinguishes the few that do.

  • From hype and predictions to impact
    Why accurate models are not enough—and how to design systems that actually change decisions and outcomes.

  • Designing human-AI decision systems
    How people actually use AI systems in practice, and what this implies for system design.

  • Evaluating AI systems in the real world
    Moving beyond accuracy and AUC to measuring whether systems improve outcomes.
  • Responsible AI beyond principles
    What fairness, accountability, and governance look like in deployed systems—not just guidelines and slide decks (the thoughts and prayers of AI)

Who is the audience?

  • Government and policy leaders
  • Nonprofits and foundations
  • Industry teams working on public sector problems
  • Academic and technical audiences interested in applied impact

Speaker Bio

Rayid Ghani is a Distinguished Career Professor at Carnegie Mellon University, with joint appointments in the Machine Learning Department and Heinz College of Information Systems and Public Policy. He leads the Data Science for Social Good Lab and co-leads CMU’s Responsible AI Initiative.

His work focuses on designing, deploying, and evaluating AI systems used by governments and organizations to improve real-world outcomes in areas such as public health, human services, criminal justice, and workforce development. He works closely with public-sector partners to move beyond dashboards, demos, and models to systems that are actually used and (sometimes) measurably improve decisions.

Before Carnegie Mellon, he was a faculty member at the University of Chicago, founded the Center for Data Science and Public Policy, served as Chief Scientist for the Obama 2012 campaign, and led Machine Learning and AI research at Accenture Labs. He advises governments on AI governance and evaluation, and has testified before the U.S. Senate and Congress on responsible AI and bias.