Artificial muscles (AMs), which encompass materials or devices capable of replicating the functions of natural muscles, have garnered significant attention in recent years, driven by the advent of various materials (advanced hydrogels, pneumatic AMs, dielectric elastomers, etc.) that exhibit exceptional properties and devices that demonstrate remarkable performance. The immense potential of AMs spans numerous industries and aspects of daily life, necessitating accelerated research efforts to meet the increasing demand. This article focuses on dielectric responsive elastomers, which are key materials within the field of AMs, highlighting advancements in theory, materials, and devices. To expedite the research and development of dielectric elastomer AM materials and beyond, we propose leveraging artificial intelligence tools to transform the artificial intelligence muscle research paradigm. Establishing an AM material database is highly valuable, as seemingly minor material data can be correlated with descriptors and target values via machine learning. Through material data mining integrating materials science and data science, we can predict potential breakthroughs in AM materials. A data-driven experimental research approach significantly reduces the number of experiments required for AM development, leading to cost savings and increased research efficiency.