Reliable AI: Successes, Challenges, and Limitations

Explore the mathematical perspective on the reliability challenges in artificial intelligence through this 52-minute lecture by Gitta Kutyniok at the Erwin Schrödinger International Institute for Mathematics and Physics. Gain insights into the current wave of AI and its unprecedented impact on industry, public life, and sciences. Understand the importance of reliability in AI, highlighted by political regularization efforts like the EU AI Act and G7 Hiroshima AI Process. Delve into recent advances in generalization guarantees and explainability, with applications in imaging problems and learning physical laws. Examine fundamental limitations in AI reliability and discover surprising connections to emerging computing paradigms such as neuromorphic and quantum computing.

Gitta Kutyniok - Reliable AI: Successes, Challenges, and Limitations