Building Truth in AI and Data

Episode 9: Martin Leach, PhD — Building Truth in AI and Data in Federal Agencies

In this episode of Practical AI in Healthcare, hosts Dr. Steven Labkoff and Dr. Leon Rozenblit speak with Dr. Martin Leach, Chief Data Officer at Black Canyon Consulting, whose team supports multiple high-impact projects in a wide array of public and private organizations.

Dr. Leach brings decades of experience from the worlds of pharma, biotech, and informatics — including leadership roles at Merck, Alexion, Biogen, AstraZeneca, and as the inaugural CIO of the Broad Institute of MIT and Harvard. Now advising public and private agencies, he’s helping to modernize how data and AI drive scientific insight across some of the world’s most complex biomedical datasets.

The discussion covers:

• AI at scale in government research – from the world’s largest sequence archives to case-report text mining in public agencies
• Truth and verification in AI – the effort to create gold-standard datasets that validate model accuracy and build public trust.
• Hybrid architectures – combining LLMs with structured APIs to ensure models return verifiable data instead of probabilistic guesses.
• AI literacy and context – why users must understand how acronyms like “SAD” can lead to misinterpretation in clinical settings.
• The next frontier: quantum AI – how exponential scaling and lower energy demand may soon reshape how models are trained and deployed.

Leach also discusses the thrill of standing up an H100 GPU cluster for under $50 to run high-end modeling.

It’s a conversation that connects science, policy, and technology, spotlighting how federal innovation in AI could ripple through clinical research and patient care in the years ahead.