Equations in Action

This presentation explores the evolution of Artificial Intelligence from ancient analog devices to modern machine learning algorithms. It covers early computational tools, the development of foundational algorithms like the Perceptron and backpropagation, and the rise of AI due to advances in data availability and computing power. Real-world applications are highlighted, including AI in board games, black hole imaging, and healthcare. The talk concludes with reflections on the lack of a unifying mathematical theory for machine learning and the importance of building such a foundation to better understand AI's capabilities and limitations.

Head and neck tumors often spread via the lymphatic system, which is why the neck region of affected patients is often treated with radiation therapy as a precautionary measure and over a large area. However, based on detailed data and customized models, we can now predict more precisely for individual patients the risk of metastases. In a clinical trial, we are using this estimation to study whether we can reduce side effects and improve the treatment effect by irradiating smaller parts of the neck in a more targeted manner.