How is AI changing the science of prediction

Hi Jorge, wellcome to CuriousAI.net. Thanks for sharing this podcast — it was a really insightful listen. Emmanuel Candès dives into some fascinating aspects of how modern statistics intersects with AI. I found his discussion on compressed sensing particularly interesting.

Compressed sensing is a signal processing technique that allows you to reconstruct a signal or image from far fewer samples or measurements than traditionally required — without losing essential information.

Candès was one of the pioneers of this field. His work laid the mathematical foundation for compressed sensing, combining:

• Linear algebra (sensing matrices),

• Optimization (convex programming),

• And probability theory (random measurements).

Compressed sensing is important in AI and data science, because implies:

• Efficiency: It enables faster data acquisition, which is crucial in resource-limited scenarios (e.g., medical imaging, astronomy, sensors).

• Lower Cost: Reduces the amount of data that needs to be stored, transmitted, or processed.

• Better Performance: In AI, using fewer, more informative features or inputs can improve learning efficiency and generalization.

One quote that really like me was when he said, “Statistics is the language of uncertainty.” It perfectly captures the idea that we need rigorous tools to make decisions in ambiguous or data-driven environments — which is central to the way intelligent systems operate today.

I’d definitely recommend this podcast to anyone looking to deepen their understanding of how AI, statistics, and data science converge from both an academic and practical perspective.