AI is on the cusp of a revolution, driven by the convergence of several breakthroughs. These include large language models that can reason like humans, and the integration of natural language and computer vision through vision-language models (VLMs). In this keynote talk, Professor Trevor Darrell will present his group’s groundbreaking work, including methods for training vision models when labeled data is unavailable and techniques that enable robots to determine appropriate actions in novel situations. Much of Darrell’s work aims to overcome obstacles—such as massive memory and compute requirements—that limit the practical applications of state-of-the-art models. He will discuss approaches to making VLMs smaller and more efficient while retaining accuracy. He will also show how LLMs can be used as visual reasoning coordinators, overseeing the use of multiple task-specific models to enable superior performance.