Abstract

This paper introduces Artificial Research by Application as one of the foundational models for automating scientific research through Specific Artificial Intelligences (SAIs), complementing the model of Artificial Research by Deduction, which is developed in a separate series of papers. It examines the architecture of SAIs designed for Artificial Research by Application, structured across three developmental stages: application (data acquisition and categorization), replication (psychological emulation and hypothesis generation), and auto-replication (self-improvement at both software and robotic levels). The paper offers detailed theoretical examples in fields such as medicine and astronomy, illustrating how SAIs might identify and respond to previously unknown phenomena. These systems are presented as preparatory frameworks for the future construction of a Global Artificial Intelligence (GAI), capable of integrating diverse scientific, economic, industrial, and administrative SAIs to conduct multidisciplinary research and make autonomous decisions. Together, the models of Artificial Research by Application and Artificial Research by Deduction are positioned not merely as technological innovations, but as philosophical milestones—advancing the shift toward AI-led epistemology. The paper concludes by reflecting on the broader philosophical and practical implications of these developments and their role in accelerating the emergence of GAI as a self-improving, rational agent of scientific discovery.