Based on the article "Does Artificial Intelligence Really Exist? Some Clarifications Around a Controversial Science" by Jean-Sébastien Vayre and Gérald Gaglio, published in Diogène (2020, No. 269-270, pp. 107–120), by Presses Universitaires de France.
Article Summary
In this article, sociologists Jean-Sébastien Vayre and Gérald Gaglio revisit the evolution of artificial intelligence (AI) as a scientific discipline. They provide a socio-historical analysis that helps explain its current development and the many technological applications that have emerged from it. Although AI became established as a scientific discipline in the 1950s, the term itself was used less frequently between the 1980s and 2010s. Nevertheless, many technological innovations developed during this period were already based on knowledge and methodologies originating from AI research.
The Birth of the Discipline
Artificial Intelligence is a relatively young discipline. It was officially founded in 1956 during the Dartmouth Conference and was initially associated with the field of computer science. Its founding hypothesis was as follows: "Every aspect of learning or any other characteristic of intelligence can, in principle, be described so precisely that a machine can be made to simulate it." AI is a deeply interdisciplinary science, at the intersection of mathematics, computer science, management sciences, and the social sciences. It is also an experimental field that seeks to define intelligence and determine whether it can be reproduced through artificial systems such as computer programs.
From Social Sciences to Neuroscience
To test their hypotheses, AI researchers have relied on concepts borrowed from a variety of disciplines. Proponents of the symbolic approach, such as Herbert A. Simon, drew heavily on concepts from sociology and cognitive sciences to understand the mechanisms of intelligence. From the 2000s onward, research increasingly shifted toward a connectionist approach. Deep learning methods gained prominence and were largely inspired by neuroscience and the functioning of neural networks.
The Renewed Interest in Artificial Intelligence
Beginning in the late 1980s, the term "Artificial Intelligence" gradually fell out of favor. This decline was largely due to the gap between the commercial expectations of the economic and political actors funding research and the actual results achieved. Since the 2010s, however, artificial intelligence has experienced a remarkable resurgence. While it is often portrayed as a technological revolution, the authors emphasize that AI research never truly stopped. Instead, it continued to evolve under different names and within other research domains. As a result, many technologies associated with the Information and Communication Technology (ICT) revolution of the 1990s—including personal computing and the Internet—were built upon software and knowledge derived from artificial intelligence research.
Conclusion
The current rise of artificial intelligence is not a sudden breakthrough, but rather the culmination of decades of research and scientific progress. Understanding the history of the discipline provides a more nuanced perspective on its challenges, its real contributions, and the limitations of the often exaggerated narratives surrounding it.