The slide from lofty promises to commercial imperatives

AI sees humans as rational economic actors, and its early promises of benefiting humanity quickly switched to creating marketable products.

In 1956 a group of mathematicians gathered for the Dartmouth Summer Research Project on Artificial Intelligence, the workshop that established the term we now use to describe systems that simulate the operations of the human mind. John McCarthy came up with ‘artificial intelligence’ to differentiate their field from that of Norbert Wiener and the cyberneticians, who were then attracting attention, and funding, for the automation of industrial processes. Unlike the cyberneticians, who were well-versed in ancient philosophy and the life sciences, the Dartmouth attendees were in part inspired by neoliberal economic theory. They shared ‘the motivating presumption that the mind was an orderly thing; that it lived inside an individual’s brain; and that it followed an implicit, reliable “logic” that could be convincingly modelled with modes of computation derived from the observation of social events’.

AI methodology took inspiration from orthodox economics, particularly by extrapolating human behaviour from a model based on a rational, calculating individual. Herbert Simon, one of its pioneers, was himself an economist and drew on Adam Smith’s studies of administration and decision-making processes to help shape what would become artificial intelligence’s ‘symbolic paradigm’: the design of systems grounded in series of decision-rules devised by specialists.

Psychologist Frank Rosenblatt found inspiration for his ‘perceptron’ – an early ancestor of ‘neural networks’ and emblematic of the ‘connectionist paradigm’ – in Friedrich Hayek’s work on market structures, which emphasised decentralised, spontaneous connections. According to this model, artificial intelligence should create a natural order capable of statistically organising the world more efficiently, functionally and rationally than individuals or collective entities such as states.

Economics and AI, fields often seen as at odds within computer science, in fact both stem from the same axioms. As philosopher Mathieu Triclot (...)

Full article: 1 575 words.