Machine Learning (ML) has historically been associated with Artificial Intelligence (AI) but has developed into an independent discipline. This paper argues for the ontological independence of ML, driven by its unique methodologies, applications, and ethical considerations. A bibliometric analysis reveals that ML research output (494,572 publications from 2017–2023) surpasses AI (283,762 publications) by 74%, reflecting its rapid growth and specialization. Unlike AI’s pursuit of general intelligence and symbolic reasoning, ML focuses on data-driven performance optimization, with impactful applications in computer vision, natural language processing (NLP), and autonomous systems. The study highlights ethical challenges—such as addressing algorithmic bias (50 occurrences), fairness (2,778 publications), and environmental sustainability (283 related works)—which emphasize the need for dedicated ethical frameworks tailored to ML. These findings propose a conceptual and practical separation between ML and AI to enable targeted research, interdisciplinary collaboration, and solutions to challenges like explainability, transparency, and sustainability. The paper underscores the importance of recognizing ML’s independence in advancing both fields.
Breaking Away From AI: The Ontological and Ethical Evolution of Machine Learning
Incremona A.;Pozzi A.;
2025-01-01
Abstract
Machine Learning (ML) has historically been associated with Artificial Intelligence (AI) but has developed into an independent discipline. This paper argues for the ontological independence of ML, driven by its unique methodologies, applications, and ethical considerations. A bibliometric analysis reveals that ML research output (494,572 publications from 2017–2023) surpasses AI (283,762 publications) by 74%, reflecting its rapid growth and specialization. Unlike AI’s pursuit of general intelligence and symbolic reasoning, ML focuses on data-driven performance optimization, with impactful applications in computer vision, natural language processing (NLP), and autonomous systems. The study highlights ethical challenges—such as addressing algorithmic bias (50 occurrences), fairness (2,778 publications), and environmental sustainability (283 related works)—which emphasize the need for dedicated ethical frameworks tailored to ML. These findings propose a conceptual and practical separation between ML and AI to enable targeted research, interdisciplinary collaboration, and solutions to challenges like explainability, transparency, and sustainability. The paper underscores the importance of recognizing ML’s independence in advancing both fields.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
