A Unified Taxonomy of 19 AI System Types: Architecture, Capability, and Deployment Analysis for the Modern AI Landscape

Abstract

This paper presents a comprehensive taxonomy of 19 distinct AI system types that characterize the modern artificial intelligence landscape. Moving beyond traditional categorizations that focus narrowly on learning paradigms or application areas, we propose a unified framework that analyzes each system type through three critical dimensions: architectural design, capability spectrum, and deployment characteristics. Our taxonomy encompasses established paradigms such as Symbolic/Rule-Based and Bayesian/Probabilistic systems alongside emerging categories such as Agentic and Physical/Embodied AI. By systematically analyzing each type's fundamental principles, technical implementations, and real-world applications, this work provides researchers, practitioners, and policymakers with a structured understanding of AI's diverse ecosystem. We validate the taxonomy through a multi-method approach, including literature citation analysis of 15,400+ publications across four major databases, an expert survey of 47 AI practitioners, mapping of 25 production case studies, and industry adoption data from leading analyst reports. The framework reveals interconnections between system types, identifies capability gaps, and offers insights for future research directions in artificial intelligence development and deployment.

Keywords