For the last 20 months, we have been working on a project funded by Western University to develop publicly available AI education resources (see https://ai-ed.ca). Through this work, we have become more aware of the nature of recent AI and its heavy reliance on mathematical algorithms from a variety of mathematical fields (such as linear algebra, calculus, probability and statistics, optimization theory, and graph theory). At the same time, we notice that educational jurisdictions are increasingly integrating AI education expectations in their curricula in ways that appear to be primarily focused on the impact and ethics of AI, with much less focus on the underlying conceptual knowledge that makes AI work. The few jurisdictions that have started to focus on the latter, in upper secondary school curricula, have realized the need for addressing mathematics concepts used in AI development. Our project’s AI education focus spans grades 1-12 and beyond, and we are especially interested in finding ways to develop AI mathematics resources that have a low floor and a high ceiling, which may engage students across a wide span of grades. In this presentation, we will elaborate on the above, share some ideas for low-floor, high-ceiling AI mathematics resources, and more generally (in the context of comparing mathematics of ancient Babylon to ChatGPT mathematics knowledge) initiate a discussion of mathematics learning in an era of machine learning. Bio: George Gadanidis is professor of mathematics education at Western University, with research interests in mathematics, technology, and the arts. [see https://imaginethis.ca] Jonathan Tan is a PhD student at Western University, specializing in AI. His research involves exploring graph (nodes connected by edges) representations of games and training agents to route and play games using graphs.