Explore a 58-minute conference talk by Petros Koumoutsakos of Harvard University on combining artificial intelligence and scientific computing for fluid mechanics. Delve into the speaker's perspective on creating alloys of AI and simulations to predict and control complex flow systems. Discover novel algorithms for learning Effective Dynamics (LED) of complex flows and the fusion of multi-agent reinforcement learning with scientific computing (SciMARL) for modeling and controlling flow-structure interactions. Examine the successes and failures in this field, and understand why integrating fluid mechanics knowledge with AI expertise is crucial for advancing scientific frontiers. Recorded on January 11, 2023, as part of IPAM's Explainable AI for the Sciences: Towards Novel Insights Workshop.