Development and Deployment of Large-Scale AI Systems and Tasks They Still Struggle With

This seminar talk by Igor Molybog, PhD, Assistant Professor at the University of Hawaii at Manoa, explores cutting-edge research on large language models (LLMs) and their multi-modal extensions. Discover the complete end-to-end pipeline for developing and deploying large-scale AI systems, with particular emphasis on the challenges and costs of data generation during advanced training stages. Learn about effective strategies for combining multiple models into inference systems to boost efficiency. Examine specific tasks where even the most sophisticated multi-modal models continue to struggle, along with approaches that have demonstrated significant performance improvements. The speaker shares insights from his research group and his experience working at Google DeepMind and Meta AI, covering topics including expanding AI's impact through novel use cases, multimodal modeling integration, and addressing core machine learning scaling challenges.