The organism–machine analogy

The organism–machine analogy is a complex figure of thought which knows many variants in science, technology and philosophy. This part traces its progression in four historical positions without imposing a periodisation: iatromechanics, developmental mechanics, cybernetics, and organology. Iatromechanics was a branch of early modern medicine that considered the body of living beings as a mechanism, and described its parts as mechanical devices (Borelli 1680). Developmental mechanics, known as Entwicklungsmechanik in German, argued that embryos are subjected to the laws of mechanics and, therefore, ontogenesis can be experimentally modified following such laws (Roux 1895; Lenoir 1989). Cybernetics was grounded on a speculative postulate that machines can imitate the key principles of living beings, namely self-organisation, self-adaptation, self-generation and self-repair. It had the ambition of implementing the structure of the organism into a teleo-mechanism (Rosenblueth et al. 1943; Wiener 1948). Canguilhem (1952/2022) developed the concept of organology to stress that technology is an extension of living beings; therefore, the machine cannot be considered as an independent model for the organism, and rather the other way around is the case. These positions show that the history of the notions of machine and organism cannot be fully separated and that they constitute together a trans-systemic paradigm.

The organism–machine analogy is a field of contrasting worldviews. The concept of organism played a different role at the time of German idealism, Weimar culture, US cybernetics, French philosophy, and feminist debates. As already seen, for Kant, the organism was a philosophical concept to emancipate natural philosophy (and the bourgeoisie) from theology. During the Weimar Republic, the organism incarnated a cultural reaction to industrial alienation and a ’call to Wholeness’ against the fragmentation of the German nation (Harrington 1996). For cyberneticians, the organism grounded a modernist project of industrial automation against the background of new forms of social autonomy that animated Western society after WWII (Pasquinelli 2023). For Canguilhem, organology was a critical response to the new mechanistic worldview of cybernetics, at the same time as French structuralism was absorbing information theory as its episteme (Wolfe 2015). In the feminist debate of the 1970–80 s (Merchant 1980; Federici and Fortunati 1984; Federici 2004, et al.), the organism became an ideal subjectivity resisting modern sciences’ rule over women’s body, but it also represented an essentialist position to be abolished as in the programmatic myth of the cyborg (Haraway 1985).

In the development of AI, the organism–machine analogy played a key role that is different from the influence of the Turing Machine and its formalisation of language. As Keller (2008) noted, the automation of the principle of self-organisation was central to cybernetics and one can better understand early neural network research following this intuition over other principles of computation. Proceeding from holistic neurology, the idea of the brain as a system of neurons capable of self-organisation, self-adaptation and self-repair influenced Frank Rosenblatt (1958, 1962) in his invention of the neural network Perceptron, which is the prototype of the current deep learning architecture. Amongst other ideas, Rosenblatt attempted to implement in the Perceptron the basic Hebbian rule of neuroplasticity (Hebb 1949), usually paraphrased as ‘neurons that fire together, wire together.’ In the early 1960s, the US Army sponsored a series of conferences and research projects on the ‘principles of self-organisation’ that gave momentum to the idea of artificial neural networks as an alternative way to computation. These debates inspired also the conceptualisation of Arpanet, progenitor of the Internet, as a communication network that could reorganise itself in case of structural failures and external attacks, as biological neural networks do (Pasquinelli 2023: 149). The principle of self-organisation was in origin a morphological principle of adaptation rather than a logical one, but cyberneticians believed that discrete-state machines, such as digital computers, could nevertheless imitate it (Ashby 1947).

Because of its technomorphic drive to equate animals, humans and machines, cybernetics should be considered first a doctrine of social governance rather than a technical discipline (Wiener 1950 became aware of this). It is not difficult to see the cybernetic agenda to automate adaptation and learning as a technocratic view of society. Already in 1951, the political scientist Karl Deutsch (1951) described advanced societies as self-organising learning networks, an analogy in which the distinction between machine and society disappeared. Although quite reductionist, Deutsch’s view is prophetic of the current predicament in which the self-organising learning networks of AI are leading the automation of labour. Whilst cybernetics was keen to describe the machine as an organism or self-organising society, it was actually an ideological view implying society had to be organised as a machine. The principle of self-organisation that manifests itself in cybernetics and in current AI is not the imitation of a feature of the living, a principle of autonomy: rather, it is a principle of automation, that is of monopoly and control over the social bond.