The philosopher Alfred North Whitehead¹ once observed: ‘Civilisation advances by extending the number of important operations we can perform without thinking of them.’ Whitehead argued that automating simpler, repetitive tasks frees up mental space for more complex thinking, facilitating the progress of human intellect and civilisation. Although often misunderstood and taken out of context, this observation faces a revealing counterpoint in the age of AI. While it is true that AI is automating and taking control of many tasks, diminishing our awareness of tasks that once required active thinking (from translating texts to handling phone calls and even programming), it also makes us more aware of our limitations and forces us to reflect on our current value in the job market. The question that arises is no longer the science fiction-like ‘Is Artificial Intelligence real intelligence?’, but the more prosaic ‘Will AI put me out of a job?’ We find ourselves in a historical dialectic that is merciless to human self-esteem: as AI progresses, we grow not less but more aware of the abilities that distinguish us from machines. Rather than thinking less, as Whitehead suggested, we end up thinking a little too much about which of our skills will be mapped, measured and mechanised by the algorithm of a technological monopoly. Techno-enthusiasts see this process of automation as a virtuous process towards social emancipation. This essay does not speculate on the implications of possible mass automation, but rather on an aspect that is often overlooked, namely the role of technology as an implicit metric of humanity.

AI is not a manifestation of superintelligence, as certain techno-enthusiast and techno-apocalyptic popular beliefs would claim.² On the contrary, it represents a mechanisation of the average intelligence of a given society. Mathematically speaking, AI works on the basis of a statistical representation (admittedly very extensive, but still statistical) of human culture as codified in the form of digital archives (so-called training datasets). AI systems such as ChatGPT process enormous amounts of this data – text, images and more – and then generate predictions and classifications based on the average values of these digital archives of collective knowledge. In a sense, it is a statistical model of the average human being, purged of extremes and anomalous behaviour. The inherent power of normalisation in AI has already been discussed by many scholars, who have recalled the origins of machine learning in the various calculation techniques of early statistics (correlation, standard deviation, logistic regression, factor analysis, etc.), emphasising the discriminatory use that was often made of these calculations in controlling social behaviour in the late 19th century.³ The normalising vocation of early statistics is also reflected and recognised in contemporary AI.

Whereas we might think that Silicon Valley computer engineers are testing AI to see if it is intelligent or not, in a dramatic reversal, it is actually AI that is being established as a collective test of our intelligence, our work skills and our standards of behaviour. As a statistical average and monopoly of the same, AI highlights what skills we possess or no longer possess as subjects in a highly automated society. AI ultimately presents itself as a reverse mirror, an inverted identity, of our social roles: what it is capable of is precisely what we no longer consider to be art or a mark of human distinction. AI, therefore, serves not only as a tool for the automation of work, but also as a yardstick, a metric of the collective workforce. AI is an implicit judgement of human skills, a managerial vision of culture in general, and its role in society is marked by this normalising power, which, following Michel Foucault, we could say has passed from the scientific institutions of modern nation states to the technological monopolies of the present.

The quantification of the world is a central practice of the Western tradition: not only for the birth of the modern scientific spirit, but above all for the developments of colonialism and capitalism, when natural resources and workers (including slaves) were meticulously measured in order to be exploited and distributed along trade routes. In the post-industrial era, this process of measurement shifted from natural resources and manual labour to mental labour and, finally, to labour as a universal capacity, to social cooperation as a productive force. Several economists have highlighted the role of knowledge in the economy, what Ricardian socialists and more recently workerists have called the general intellect.⁴ Although knowledge has always played an economic role, even in ancient civilisations (needless to say), and despite the acceleration of this process more recently with the rise of information technologies, AI is a further step in the development of knowledge technologies. For the first time, AI represents a mechanisation of general intellect as a collective form and a monopoly of this same collective dimension on a global scale. What truly makes AI superhuman is that it embodies culture as a collective actor (as a Gesamtarbeiter, or ‘general worker’, as Marx would say).

As noted, AI does not merely automate an extraordinary number of tasks, it also plays an implicit and explicit role in the measurement of these tasks. To be precise, it is establishing itself as an analytical metric for evaluating collective labour and intellect. To understand how AI operates as a metric, we should revisit the principles that gave rise to modern computing. The path to AI began with the work of American mathematician Claude Shannon, whose invention of binary computation and information theory laid the foundation for the digital revolution beginning in the 1950s. Shannon’s theory of computation demonstrated that electromechanical devices such as relays (and later microchips) could perfectly implement Boolean logic, while his theory of information described how to measure information statistically and transmit it with minimal signal loss. Shannon’s ideas helped translate mathematical logic into binary logic, and to later digitise the main forms of mass culture and communication. AI contributes something different to the digitisation process: the automation of classification, or the ability to recognise and interpret signs and objects.