CXOTalk Top Data Scientists Explain Bad Data, Poisoned Datasets, and Other AI Killers | CXOTalk #896
44 snips
Oct 9, 2025 Join Dr. David Bray, a tech policy expert at the Stimson Center, and Dr. Anthony Scriffignano, a data science leader, as they dive into the hidden threats of bad data and poisoned datasets in AI. They discuss the Five Ms framework for identifying AI failures and why organizations often rush into AI adoption without proper vetting. Learn about the risks of generative AI, the importance of critical thinking and ethical oversight, and how to recognize malicious data campaigns that can undermine your AI systems.
AI Snips
Chapters
Books
Transcript
Episode notes
AI Needs Trustworthy Foundations
- Generative AI depends on training data as its foundation and that foundation can be unstable.
- If you ignore data provenance or missingness, your model may collapse like a castle on quicksand.
Start With Problem, Not Tool
- Do not lead with tools; start with a clear problem and mission before using AI.
- Ask how AI produces net-new value and whether the chosen tool fits the problem.
Real Errors From Bad Training Data
- David gave an example where a model incorrectly reported Georgia's population as ~350 million due to a training typo.
- He also recounted policy deployments that placed COVID resources by residence instead of workplace and misread demand.
