What Can the Theory of Cryptography Tell Us About AI Verification?

Name: What Can the Theory of Cryptography Tell Us About AI Verification?
Uploaded: 2025-04-23T20:53:10.000Z
Duration: 1 h 2 min 15 s
Channel: Shafi Goldwasser
Description: Applies interactive proof systems, cryptographic watermarking, and worst-case adversarial guarantees to verifiable machine learning.

Shafi Goldwasser

01

Mathematics and Computation: The Landscape of P vs. NP

Avi Wigderson · 1:02:49

This lecture establishes the foundational concepts of computational hardness and the P vs. NP problem, which are essential for understanding the security assumptions underlying all cryptographic systems.

02

Proofs, Knowledge, and Computation: Zero-Knowledge Interactive Systems

Silvio Micali · 1:03:26

Building on the concept of computational hardness, this lecture introduces interactive proof systems and the revolutionary idea of zero-knowledge proofs, demonstrating how interaction can verify knowledge without revealing it.

03

IP = PSPACE

Ryan O'Donnell · 1:18:47

Expanding on interactive proofs, this lecture delves into the IP = PSPACE theorem, showcasing the immense power of interactive systems and their theoretical equivalence to polynomial space computations, which is crucial for advanced cryptographic protocols.

04

The Birch-Swinnerton-Dyer Conjecture: A Millennium Prize Problem at 25

Chris Skinner · 52:16

Transitioning from theoretical complexity, this lecture introduces elliptic curves through the Birch-Swinnerton-Dyer Conjecture, providing the mathematical foundation for modern public-key cryptography, including ECC.

05

Introduction to Hash-based Post-quantum Cryptography (PQC)

Christof Paar · 1:40:59

Following the mathematical foundations of public-key cryptography, this lecture addresses the critical challenge of quantum computing by introducing hash-based post-quantum cryptography, preparing learners for future cryptographic paradigms.