Seeming Human: Artificial Intelligence and Victorian Realist Character, by Megan Ward

Abstract

Reviewed by: Seeming Human: Artificial Intelligence and Victorian Realist Character by Megan Ward Tamara Ketabgian (bio) Seeming Human: Artificial Intelligence and Victorian Realist Character, by Megan Ward; pp. viii + 181. Columbus: The Ohio State University Press, 2018, $64.95. How should one create a sense of human intelligence and subjectivity? What does it mean to think or act like a human? According to Megan Ward’s new monograph, Seeming Human: Artificial Intelligence and Victorian Realist Character, both Victorian realism and mid-twentieth-century artificial intelligence explore these questions through fictional characters that use “bald artifice” (12) and mechanism “to seem—rather than be—human” (3). Ward seeks to contest traditional realist models of depth psychology and interiority and instead to theorize alternate, “transhistorical” (3) forms of “human-likeness” and verisimilitude (2). She does so by reading forward and backward, positing “the emergence of AI in the twentieth century as at once a formal afterlife of Victorian realist character and a theory for rereading those same novels” (7). Often elegant and intriguing, these comparisons highlight the rewards of imagining subjectivity, learning, environment, and communication through the networks, feedback loops, and neural-net models of early AI before [End Page 127] the field’s later turn to intelligence as dematerialized thought. Many of Ward’s broader conclusions are not novel; she is far from the first critic to suggest that Victorian characters also happen to be highly fictional, artificial, mechanical, nonhuman, and posthuman creations. The means by which Ward makes these claims, however—through readings of the shared informatic texture of character and AI—are engaging, refreshing, and pose promising directions for Victorian studies in the future. Seeming Human bases its chapters on four limit cases of realist character—development, predictability, flatness, and mind—and pairs each with a cognate model of AI from the mid-twentieth century. Ward’s first chapter examines Norbert Wiener’s cybernetic feedback loop as a temporal structure for domestic routine and mechanically recursive character development in fiction by Charlotte Yonge, Elizabeth Gaskell, and Margaret Oliphant. This chapter, however, which repeatedly compares characters to responsive steam engines, might profit from existing scholarship by Sally Shuttleworth and Herbert Sussman on the steam governor’s close association with narratives of mechanistic learning, adaptation, and Smilesian self-help. Chapter 2 investigates predictable character as part of connected stochastic information systems in sensation and mystery fiction by Charles Dickens, Wilkie Collins, and Mary Elizabeth Braddon. Seeming Human’s third and most compelling chapter uses Alan Turing’s 1950 imitation game to propose a “new theory of flatness” in Anthony Trollope’s The Eustace Diamonds (1871) and The Prime Minister (1876) (15), through human “fakers” who expose the mimetic surfaces and simulations so integral to realist character (2). Ward’s final chapter treats the fictional characters of Thomas Hardy and Henry James as anticipating AI’s split between embodied and cognitivist models in the 1960s: between the perceptron, with its associative webs of material sensation and behavior, and the physical symbol system, defined by “abstract, immaterial cognition” (16). Although Ward does not use the term, her study is most original as a behaviorist theory of realist form. Seeming Human treats character as an inaccessible black box defined by inputs, outputs, and linkages—by material “behavior, connections, surfaces, and perceptions” rather than interior thought (10). Her study might benefit from a discussion of cybernetics as a science of life and life-likeness, since it arose along with the study of homeostatic processes in animals and organic systems. These insights could further illuminate her second chapter, by drawing links between the information flows of character and the organic social webs often allied with realist fiction. Ward’s cybernetic formalist-historical approach has an early precedent in Hugh Kenner’s The Counterfeiters (1968), an eccentric study of eighteenth-century and modernist literature through Turing, Wiener, and other AI figures devoted to counterfeiting humans. While unremarked by Ward, The Counterfeiters usefully amplifies Seeming Human by stressing a perpetual occupational hazard of mechanical fakery: the risk of tautology and self-parody. For, as Kenner suggests, “[t]he computer simulates thought when thought has been defined in a computer’s way” (The Counterfeiters [Indiana University Press, 1968], 40...