Definitions are tools with heuristic utility. A thin technological definition of AI may be intelligence displayed or simulated by code or machines. In this sense, a basic general definition of AI in the spirit of the 1956 Dartmouth workshop that emphasizes the aim to simulate human intelligence in computers is useful.

Definitions of AI abound depending on what you intend to use AI for. None are wrong. All are useful. But a tool, like a nail, can never be defined in isolation. Tools exist in relation to the interminglings that they make possible, or that make them possible, according to Deleuze and Guattari in “A Thousand Plateaus”, (1980, p.90).

Beyond computer science definitions of AI that emphasize code, what is definite is that AI is a social reality. This takes us outside a preoccupation with technologically driven forms of thought that can produce generalizations. What we need to understand is how AI as a social reality takes shape, what it does in society, and what cultural, political, economic, and social processes of power it releases and subordinates.

Unravelling which unfreedoms AI removes, how it releases the imagination, and what prejudices and prejudgements it transports from the European enlightenment into the future, its rhizomes, tentacles, and praxis ought to be our primary concern.

This requires a deeper characterization of AI, that highlights a broader constellation of underpinnings – one that underscores how AI reaches out and intertwines people and their lives in society. This allows us to interrogate AI in society and its politics in a world migrating towards increasing algorithmically-driven digital capitalism, or what Shoshanna Zuboff (2019) calls “surveillance capitalism.”

In this regard, AI becomes a multiplicity and not just one tool. Rather than one technology, AI is a collection of different components, including algorithms, that compiles itself into an assemblage of technological components. AI is a suite of technologies and tools that is a ubiquitous apparatus that has a social and political ideological undercarriage.

Unlike the “back end” of our present technical architecture, AI consists of the liaisons or relations among several parts which are themselves unfinished and contested. At the core of the digital society under construction are the deep human and material roots of AI’s complex assemblages. AI is thus a multiplicity that sees the social world not in terms of fragments or individuals, but as an abstract collective, a collectivity, or a complex agglomeration of different components.

Because we cannot see inside AI’s complex assemblages, including individual smartphones and tablets, data lakes, data cooperatives, data refineries, data embassies, and subscription services to server farms, we do not grasp it as a ubiquitous apparatus that entangles our lived experiences and life worlds.

We are further unable to see the programmers, the designers, and the chip manufacturers as forces in their own right. Moreover, we cannot see all the people affected by AI as separate from AI. AI is both the medium to and a message about (and even from) the future, according to Amanda Lagerkvist.

At the material level, AI co-constitutes the lived social and political reality of citizens as it gives identity to humans as subjects in relations of precarity, prejudice, failure, inequality, and success. AI models now affect people at critical life moments at interviews, applying for credit, and work finding the talent it needs. AI’s components now function as an assemblage that produces incorporeal consequences that are attributed to selection panels and other bodies.

AI is not human. It has no self-reflective capabilities and does not pray or hope. It is a human creation that reflects human-designed processes on human-created machines. While it is an adjunct to our perceptions and thoughts, we have yet to decide on how we will reconcile AI with the ideas of human dignity and autonomy as it transforms the strategies and doctrines we develop and deploy.

When humans use AI-enabled network platforms to generate an advertising campaign or to design a fashion collection, they benefit from an assemblage of technological components and a type of data scraping and distillation that no generation before them has ever experienced.

These assemblages of tools and the interminglings that they make possible open a portal into a form of human-machine dialogue that has never existed before. In this regard, AI can mould human activity in ways not yet fully understood – or are precisely definable or expressible – by the human user. This provokes the uneasy questions:  With what objective function is such an assemblage operating? And by whose design, and within which set of regulatory parameters? The answers to these social and political questions will continue to shape lives and societies over the next decade.

Manuel DeLanda observes that an animal is defined less by its organs, species, and genus than by the assemblages into which it enters. Spiders are not insects. They are arachnids.

Likewise, AI is not defined by its type, technological form, code, and algorithms, as it is defined by the layered socio-political settings where it functions, and comes into contact with humans. DeLanda argues that society itself is a palimpsest of assemblages. But all of our social assemblages are now being reformatted by AI.