AI on the path to good decisions

Abstract

This article examines objections to the use of artificial intelligence (AI) in decisions affecting humans and argues that most objections rest on anthropocentric assumptions rather than evidence about decision quality. Critics often presume that human decision makers are uniquely capable of moral and contextual judgment, while AI systems are inferior, opaque or hostile. The article challenges this view arguing that modern AI systems are an expression of collective human intelligence and ethics, built on the image of human cognition, trained on curated human decisions and operating on best decision making frameworks. Drawing on empirical studies the article shows that modern AI systems offer unique advantages in decision-making, thus may already in some domains be as good at making good decisions as an average individual human. The article deconstructs main criticisms of AI decision-making and introduces the novel argument of inseparability between human and AI decisions. Human and AI contributions are increasingly intertwined, AI involvement is latent and appropriated by humans, making existing accountability frameworks based on a clear human–AI boundary obsolete. The article advocates for the development of agnostic decision-making frameworks that apply universal accountability to both human and non-human agents and provide a path to better decisions.