Technological failures, controversies and the myth of AI

Pleased to have a new book chapter out in a book edited by Simon Lindgren, the leading digital sociologist:

📜 A. Ballatore & S. Natale (2023) Technological failures, controversies and the myth of AI. S. Lindgren (ed.) Handbook of Critical Studies of Artificial Intelligence. Edward Elgar. [web]

“In the popular imagination, the history of computing is often represented through a dynamic of linear progress, made of dazzling engineering inventions and scientific insights that advance technical and scientific areas. Yet, the evolution of computers and the software that animates them have followed very different paths. While hardware has seen decades of marked improvement in terms of memory and computational speed, the history of software, as noted by Nathan Ensmenger, “is full of tensions, conflicts, failures, and disappointments” (2012, p. 10). At various points in the evolution of computing, as machines became cheaper and more powerful, programming them to perform useful tasks proved costly and frustrating (Ceruzzi, 2003). Developing an approach that considers the role of failure, in this sense, is an urgent task for a critical history of digital media that takes into account the fundamental role played by software and the programming of computers and other digital tools (Balbi & Magaudda, 2018).

From the perspective of a cultural history of technological failures in computer science, artificial intelligence (AI) represents a particularly interesting case. At various times through- out the evolution of computer science, AI has been regarded as a failed technology, unable to meet the grandiose promises and goals set by the researchers and proponents of its paradigms; at other stages, including the present day, it has been shot through with considerable, perhaps excessive, enthusiasm, attracting considerable investment and attention from the techno-sci- entific world and the public sphere (Ekbia, 2008). Its tumultuous history, consequently, has been read by historians as a schizophrenic alternation of failures and triumphs (McCorduck, 1979; Crevier, 1994; Russell & Norvig, 2021).

This chapter questions this narrative in order to propose a different point of view, according to which the alleged failure of AI does not concern specific stages of the decline of its various paradigms but constitutes an inescapable element of it along its entire historical trajectory. By highlighting the functional role of skepticism and controversies between AI proponents and critics, such a perspective suggests that the construction of a technological myth around AI has been facilitated and enabled by the ongoing controversy between those who emphasize its successes and those who emphasize its failures. To understand the impact of technological failures – understood as cultural constructions rather than objective events (Gooday, 1998; Lipartito, 2003) – it is, therefore, necessary to abandon a dichotomous dimension that opposes success too rigidly to failure. The possibility of failure of techno-scientific projects represents instead a symbolic and material resource that should be considered an integral part of their development. Addressing the issue of failure in regard to AI, moreover, is particularly urgent in the ongoing phase, in which AI systems are heralded as the defining technology of our era, often with the promise of infinite possibilities.”