AI before AI

Stephen Cave, Kanta Dihal, and Sarah Dillon, eds., AI Narratives: A History of Imaginative Thinking about Intelligent Machines. New York: Oxford University Press, 2020. 424 + xiv pp.

Stephen Cave and Kanta Dihal, eds., Imagining AI: How the World Sees Intelligent Machines. New York: Oxford University Press, 2023. 423 + xviii pp.

Emergent artificial intelligence (AI) technologies seem to augur a break between poetics today and the poetics of yesteryear. Were we to ask our handheld magic mirrors what kind of poetics now matter most urgently, they (or at least many of the SubStacks they bring us) might well answer: “Today, poetics is whatever one makes of generative AI—or whatever generative AI makes of poetics.” As we strive to understand the forms, cultures, and histories of past literature, we may feel that such AI—or, to be more exact, the hype it receives—casts a chilling shadow over us by instead highlighting how in the near future writing may involve processes that at present we cannot quite comprehend. Still, as the essays in these volumes collectively demonstrate, the past still matters for the future. Whatever twenty-first-century AI technologies may bring, we are motivated to develop such technologies, orient their use, and shape their reception by fears and fantasies that remain fundamentally continuous with visions of AI that date back well before the term was invented. As the editors of Imagining AI put the point, in the pithy first sentence of their introduction: AI “was a cultural phenomenon long before it was a technological one” (3). And such AI was—and, these volumes argue, remains—a diverse phenomenon, “shaped by the particular histories, philosophies, ideologies, religions, narrative traditions, and economic structures of different countries, cultures, and peoples” (4). In this view, we need a historical and cultural poetics of AI to know from whence AI, in its diverse forms and scales, has come, where AI might be going, and what relations of power AI sustains and undermines.

The trajectory charted by these related essay collections evinces the evolution, over the past decade, of a specific institutional awareness of these historical and cultural dimensions of AI. The first of them to appear, AI Narratives: A History of Imaginative Thinking about Intelligent Machines, originated, its acknowledgments tell us, in workshops held in 2017 and 2018 by “the AI Narratives project, a joint initiative of the Leverhulme Center for the Future of Intelligence at the University of Cambridge and the Royal Society” (xiii). The history AI Narratives develops is very much a Western one, and its essays move forward along a familiar timeline. A sprightly first chapter by Genevieve Liveley and Sam Thomas details the automata and other intelligent devices depicted by Homer in book 18 of the Iliad and elsewhere. Treatments of relevant medieval and early modern figures follow, with standout chapters from E. R. Truitt, on demonic or otherwise preternatural notions of augmented or artificial intelligence, such as magic mirrors and mechanical sentries, and from Julie Park, on eighteenth-century speaking machines. The collection’s second half, about which more will be said below, then surveys “modern and contemporary” AI narratives.

Without narrowing the broad temporal sweep its predecessor volume AI Narratives covers, Imagining AI: How the World Sees Intelligent Machines globalizes the scope of the inquiry. The attention Imagining AI pays to the whole world corresponds, it seems, to a reorganization of the AI Narrative project, circa 2018, into “the Global AI Narratives (GAIN) project at the Leverhulme Centre.” The essays this volume includes emerged from twenty workshops organized there between 2018 and 2021, some of them virtual or hybrid. After an introductory chapter that usefully parses what vocabularies different languages use to grapple with artificial intelligence, these contributions are “clustered geographically” to “group chapters from similar linguistic or cultural backgrounds” (7). A third associated volume, Feminist AI: Critical Perspectives on Algorithms, Data, and Intelligent Machines, edited by Jude Browne, Stephen Cave, Eleanor Drage, and Kerry McInerney, was published at the end of 2023.

The Leverhulme Centre’s various Narrative AI projects intervene significantly in AI discourse, in ways exemplified by the second half of the AI Narratives volume, which includes such deftly constructed accounts of concepts of AI in literature, popular culture, and film as coeditor Kanta Dihal’s rich meditation on AI, slavery, and revolt (as represented in texts including Karel Capek’s play R.U.R., which first names the figure of the “robot” as such), and Kate Devlin and Olivia Belton’s insightful survey of “fembots” and other robots for sex and love. In his contribution, co-editor Steven Cave focuses on nonfiction futurism by Ray Kurzweil and others—work in which “AI appears to be a kind of Holy Grail” (309). In the context of surrounding chapters treating modern and contemporary imaginative visions of AI, Cave may seem to mark out an anomalous direction for research when he proposes that “the idea of AI-enabled transcendence has to a significant extent been transmitted by works purporting to be nonfiction” (322). But as the volume introduction cowritten by Cave avers, it is a main aim of the collection to radically recontextualize tropes such as that of transcendent AI—an especially American vision, Cave suggests—even or especially as they appear in nonfiction.

AI Narratives, its editors tell us, seeks to assemble an archive of stories about thinking machines that can give body across a full range of discourse to Sheila Jasanoff’s important but somewhat arid concept of “sociotechnical imaginaries.” In so doing, the editors explain, they seek to “highlight relationships between society, technology, and the imaginary that are not included in Jasanoff’s definition” (7). The imaginaries they would explore very much include “visions that are not collectively held, ones that destabilise institutions, and subaltern narratives” and might thereby “challenge the ‘aspirational and normative’ dimension of Jasanoff’s concept” (7) (Jasanoff 2015: 5). Imaginative AI narratives are here mobilized to subvert the dominant discourse surrounding AI, by suggesting how much of it is, if not exactly fictional, no less myth mongering for that, and by excavating what kinds of dread and desire underwrite prose about AI that purports to objectively ground itself in what is, or could be, real.

The essays in the sequel volume Imagining AI expand this historical and cultural poetics of AI with detailed accounts of how intelligent machines have functioned, or have been thought to function, around the world. Abeba Birhane’s essay “Algorithmic Colonization of Africa,” for example, brings into focus the consequences of activities in continental hot spots like “‘Sheba Valley,’ in Addis Abeba, ‘Yabacon Valley’ in Lagos, and ‘Silicon Savannah’ in Nairobi,” showing how the values tolerated there in such application areas as the digitization of lending redound to anchor injustice globally (251). A subsection on East and Southeast Asia offers deep dives into how classical Chinese philosophy (per Bing Song) and ancient Chinese texts on technological devices (per Zhang Baichun and Tian Miao) imagine kinds of AI up to several millennia before the coinage of the term. Hirofumi Katsumo and Daniel White, meanwhile, provide a chapter on artificial emotional intelligence in Japanese robotics that shows an exemplary engagement with the technical dimension of a sociotechnical imaginary.

In this regard, one might want to ask for yet another volume: one that could build on the exploration here of historical and global dimensions of imaginative thinking about intelligent machines, by more fully integrating modern technical AI discourses with their cultural doppelgangers, so that neural nets might be found in the index near neoliberalism and Neuromancer, or recommender systems between rebellion and religion. Having rediscovered AI as a cultural phenomenon and analyzed it with the tools of historical and cultural poetics, these volumes leave us poised to consider what a formalist poetics of AI might reveal about the culture of contemporary AI and, moreover, about the making of emergent AI technologies. How might narratologists chart the parameters of latent space? Can data scientists infer patterns of causation across rhyme schemes? These volumes prove so successful at sounding thematic resonances between imagined and achieved AI, and between AI’s cultural and technological zones of expression, that they suggest an exploration of formal resemblances across these domains would likewise prove immensely rewarding.

References:

Browne, Jude, Stephen Cave, Eleanor Drage, and Kerry McInerney. 2023. Feminist AI: Critical Perspectives on Algorithms, Data, and Intelligent Machines. Oxford: Oxford University Press.

Jasanoff, Sheila. 2015. “Future Imperfect: Science, Technology, and the Imagination of Modernity.” In Dreamscapes of Modernity: Sociotechnical Imaginaries and the Fabrication of Power, Sheila Jasanoff and Sang-Hyun Kim, 1–33. Chicago: University of Chicago Press.