How AI systems work by recycling a narrow set of visual tropes in their training data that have very little to do with human notions of intelligence or creativity is shown.
Abstract
This article addresses the question of the ‘magic of AI’ from two interrelated perspectives: first, as a form of belief in the seemingly magical or miraculous ability of AI to replicate human-like knowledge, intelligence and creativity; and second, as a kind of magic trick used by the tech companies to make users believe in AI's extraordinary powers. By reframing AI as both a system of belief and a ritualistic marketing practice, the article argues for a shift of focus away from abstract philosophical debates about AI's existential risks towards a more ethnographically grounded analysis of how such illusions of magic – of AI's intelligence and creativity – are produced and sustained in the first place. Drawing on an ethnographic case study of how text-to-image (TTI) AI models represent controversial historical events such as the 1984 Ethiopian famine, the article shows how AI systems work by recycling a narrow set of visual tropes in their training data that have very little to do with human notions of intelligence or creativity. Underpinning the popular discourse of AI as magic we can thus find an emerging system of power in which tech companies sell the ‘miracle’ version of AI, while users receive only the ‘magic trick’.
This article examines how practitioners engage with the technology's perceived opacity by placing practitioners’ own descriptions of AI within broader anthropological discussions of magic, and argues that references to ‘AI as magic’ help situate rapidly evolving technological systems within everyday lifeworlds.
It is argued that anthropology requires a new conceptual framework for understanding the contemporary social life of artificial intelligence (AI), and "the magic of AI" is proposed as an analytical concept that shifts attention from AI's technical capacities to the beliefs, uncertainties, and sociotechnical imaginaries through which it acquires authority and efficacy.
M. Baas, Roanne van Voorst· Anthropological Theory· 2 citations
This essay reframes the discourse around large language models (LLMs) by drawing on anthropological and linguistic theories of magic as a positive transformative force. While current discussions often cast AI as mysteriously dangerous – a form of “dark magic” that beguiles users and obscures ethical concerns – we argue that LLMs represent a fundamental shift in human-technology interaction, one requiring ritualistic linguistic precision rather than technical commands. Drawing on theories of performative language and magical rites, we propose that effective prompt construction functions as ritual magic through three essential elements: clear and specific language, contextual framing, and structured sequencing. In this, we show that AI is not a singular tool operation, but rather a mode of forming a mutual cybernetic relationship. Through case studies from the AI Anthropology Toolkit, we show how skilled practitioners harness the “magical” capabilities of LLMs for analytical insights through careful prompt engineering. We then examine implications for business anthropologists working in organizational settings, exploring how this cybernetic relationship enables distributed agency and co-becoming between human expertise and computational pattern recognition. This perspective provides both a theoretical contribution to understanding human-AI interaction and practical guidance for anthropological practice in an era of increasingly sophisticated AI collaboration.
T. Malefyt, Matt Artz· Journal of Business Anthropo...· 1 citation
A defining if under-acknowledged feature of large language models (LLMs) is their non-indexicality: what AI
engines “know” about the world reflects only their ability to predict and successfully imitate existing textual representations of
that world. Like fictional narratives, therefore, they traffic not in truth or falsity, but believability. Helpfully, literary
critics have experience thinking critically about the ways in which believability is shaped by existing ideologies and biases — an
acknowledged limitation of AI-generated texts as well. Drawing on Neilsen, Phelan and Walsh’s rhetorical model of fictionality,
this article uses Chimamanda Ngozi Adichie’s metafictional short story, “Jumping Monkey Hill” as a laboratory to explore the
relationship between fictionality and believability in a situation where rhetorical cues about a text’s ontological status are
misleading. From there, I suggest how literary ways of reading might inform our engagement with AI texts — including and
especially those that present themselves as factual.
Generative AI is relatively new, but the way it produces utterances is strikingly familiar. Plato distinguishes Socratic teaching, which aims at truth, and the modes of ‘teaching’ employed by various sophists, rhetoricians, and orators, which aim at plausibility and winning arguments. Generative AI, like these latter figures, works by sounding plausible. This paper argues that there are fruitful comparisons to be made between the rhetoricians-as-teachers portrayed by Plato, and generative AI for education. Five key areas will be considered: (1) truth versus plausibility as aims; (2) how plausibility is established by wide consensus, with implications for AI bias; (3) how the utterances of the rhetorician or AI chat interface can be seen as a form of mimesis, with consequences for what is meant by ‘conversation’; (4) Cognitive and psychic dangers when mimesis becomes an habitual mode of interaction; and (5) the hype and concomitant economic pull that comes with the arrival of the rhetorician-as-teacher and AI education tools, respectively. It is argued that these comparisons have meaningful implications for AI in education, both as a diagnosis of a current state and as a more normative steer for future developments.
Clare Jarmy· The Journal of Classics Teac...· 0 citations
Artists and cultural theorists, although they use dif ferent means, share the task of problematizing culture. The urgent task now is to critically examine the ways we use AI text-to-image generators and create space for ref lection about engaging with these systems. I approach this through what I term the cultural logic of computational capitalism, drawing on Fredric Jameson and Bernard Stiegler. The paper addresses how arts and humanities can help overcome this logic and transform AI’s visual culture itself. Such an inquiry is essential given that AI text-to-image generators not only disrupt traditional art production but also concentrate creative power in the hands of a few dominant platforms. While these concerns are global, they require specific regional responses. Focusing on East-Central European countries, I argue that the underrepresentation of their visual cultures in AI models stems from their semi-peripheral status within global technological and economic systems. Rather than simply feeding existing AI models with better regional training data, I propose supporting dissident artistic practices that promote regional digital and digitally sustainable cultures.