Mathematical Foundations of Modern AI Systems
This course develops clean mathematical models for the sources of power in modern AI systems. It examines when apparent obstacles become non-obstacles and keeps proved results, empirical evidence, plausible mechanisms, and open questions separate.
Questions first
Each lecture has a broad mathematical topic and a guiding question about why the modern recipe can work, or where its explanation stops.
Mathematics as an anchor
Positive results are paired with lower bounds, counterexamples, or assumptions that expose the boundary of the claim.
No propaganda
The course does not assume that current approaches will succeed completely, or that they must fail. The evidence decides what can be said.
Start with the syllabus, see the dated course schedule, or review the coursework and research-question note. Registered students use Canvas Assignments for submissions and Canvas Grades to view recorded marks.