The ongoing debates on defining intelligence and Artificial Intelligence (AI) reflect their complexity, and the different perspectives on new paradigms such as Hybrid AI, provide a new context to broaden the challenge of grasping them. Recognizing AI variants is the basis for applying and dealing with AI-based systems. One of these new paradigms is Hybrid AI, which combines Symbolic and Subsymbolic approaches and aims to optimize their performance by opening new applications and overcoming existing limitations. The aim of this paper is to revisit the definitions of both approaches and proposes a novel Hybrid AI paradigm (Concurrency and Semi-symbolic), which demonstrates better performance compared with purely symbolic or purely Subsymbolic approaches that are widely used in the literature.