Phase changes are observed in our everyday life in nature and many industrial applications ranging from dew condensation on insects, water droplet harvesting, electronics immersion cooling in data centers, climatology prediction, and materials manufacturing. Unveiling the thermofluidic principles that can make efficient use of phase changes in these systems has been a daunting challenge for a long time. Central to this understanding is the extraction of interpretable and rich datasets from dynamic and fast-moving liquid-vapor interfaces, represented as bubbles or droplets. These challenges may be addressed by using the vision intelligence, scientific machine learning, and data analysis. In this talk, I will showcase a key approach proposed by my group, called a vision-based framework that streamlines automatic data processing, which has been nearly impossible to achieve with conventional methodologies. Then, I will introduce examples that learn, understand, and predict the dynamic nature of phase change phenomena via AI technologies. Finally, I will end my talk by briefly discussing the novel and state-of-the-art machine learning approaches that allow us to explore previously undefined features and hidden mechanisms that would represent a game-changing innovation for thermal energy science and applications.