From Automation to Autonomy: The Evolution of Intelligent Technological Systems

Abstract

Intelligent technological systems have undergone a profound evolution over the past century, progressing from fixed automation that executes predetermined sequences toward autonomy that enables systems to perceive, decide, and act under uncertainty with diminishing human involvement. This article reviews that evolution, clarifying the conceptual distinction between automation and autonomy, tracing the historical trajectory from mechanization to contemporary autonomous and agentic systems, and synthesising the frameworks used to characterise intermediate degrees of machine independence. We argue that the shift from automation to autonomy is not merely a matter of degree but a qualitative transition: automation follows fixed rules within bounded, predictable environments, whereas autonomy entails adaptive, goal-directed behaviour in open, uncertain ones, made possible by advances in sensing, computation, machine learning, and control. Drawing on established models—including the levels-of-automation taxonomies of Sheridan and Verplank and of Parasuraman, Sheridan, and Wickens, and the widely adopted six-level SAE J3016 standard—we present a generalised levels-of-autonomy framework and a structured comparison of automation and autonomy across dimensions of adaptability, decision scope, uncertainty handling, learning, and human involvement, illustrated through figures and tables. We then examine the enabling technologies and the domains in which the transition is most advanced, from self-driving vehicles and smart manufacturing to autonomous laboratories and software agents, before analysing the challenges that accompany increasing autonomy: trust and verification, accountability and liability, safety and predictability, human–autonomy teaming, and the persistent gap between the rhetoric and the reality of “full” autonomy. We conclude that the foreseeable future is not one of wholesale human replacement but of calibrated human–autonomy collaboration, and that the central challenge is designing systems whose autonomy is matched by trustworthiness, transparency, and appropriate human oversight.