Open algorithms: an interdisciplinary inquiry of artificial intelligence systems

Abstract

We live in the “Age of Algorithms,” where algorithms shape human behavior and social structures, influencing power dynamics, financial growth, and how we interact amongst us. This “Algorithm Era” presents opportunities and challenges that must be addressed. Aiming to mitigate algorithms' harmful effects on society, we argue that an inquiry is necessary. Our inquiry revisits and characterizes the classical understanding of algorithms to name them ‘closed algorithms.’ Our research theoretical evidence suggests the emergence of ‘new beasts’ in the realm of algorithm studies, which we are naming ‘open algorithms.’ They require enormous energy capacity, infinite user interactions, massive data available, and cutting-edge hardware and software capabilities, playing a far more active and decisive role in society than ever. The dichotomy between ‘closed’ and ‘open’ algorithms is evident when we compare the classical algorithm system computer science studies for around 100 years with modern AI agents. Open algorithm’s behavior and outputs are challenging to control. It is hard to explain their logic and reasons solely through examining their code (as is a typical approach when dealing with ‘closed algorithms’). This doctoral research designs an inquiry based on open algorithms' epistemology, ontology, and ethics aspects. This raises the following questions: What are these new beasts in the realm of algorithms that we call open algorithms? What makes open algorithms distinct from closed algorithms? What knowledge is necessary to know, recognize, and understand open algorithms? At first, we share our rationale for choosing the terms ‘open’ and ‘closed’ algorithms. Then, we overview state-of-the-art knowledge, seeking to explore the origins and foundations of ‘open algorithms.’ We conduct exploratory dialogues and critical algorithm studies with top-notch specialists from various fields (e.g., computer science, philosophy, political science, engineering, design, sociology, and the Law). With Pragmatism Philosophy lenses, we inquiry open algorithms. The unexpected nature of the object outcomes suggests controllability, scalability, and the knowledge involved in its use are issues to be considered. Some first steps towards grasping open algorithms are understanding their ‘use’ (in loco), their ‘oneness’ (people, process, artifacts), and their constituents (energy, hardware, software, agents, data, interface, and society).