There are hidden patterns and discoveries yet to be made in existing astronomical datasets. With astronomy facing an influx of petabytes of data from advanced instrumentation, the limitations of traditional analytical methods are becoming more evident and urgent. Artificial Intelligence (AI) presents a pivotal opportunity; however, astronomy poses challenges to the current generation of AI methods, primarily adapted for different types of datasets and goals. Specific challenges include: 1) the complex nature of astronomical data—heteroskedastic, multivariate, sporadically sampled, and unannotated; and 2) the task of extracting meaningful physical insights. In this talk, I will discuss the need for developing AI for astronomy. I will delve into the perspectives, challenges, and opportunities ahead, emphasizing the necessity of preparedness as we navigate this evolving landscape.