Would you describe AI as a commodity?

I re-read Appadurai's brilliant essay 'Introduction: Commodities and the Politics of Value' or "The Social Life of Things", attempting to apply its principles to AI. The complexity of the essay and my lack of practice in reading academic texts made re-reading it a challenging but certainly interesting experience.

Let's start with a simple concept:

Economic exchange creates value. Value is embodied in commodities that are exchanged...what creates the link between exchange and value is politics, constructed broadly [1]

Therefore, every object that becomes a commodity requires an examination of its social context, since in becoming a commodity it inevitably takes on a social role.

But exactly how does an object become a commodity?

Commodities are things with a particular type of social potential, that they are distinguishable from "products," "objects," "goods," "artifacts," and other sort of things – but only in certain respects and from a certain point of view. [2]

Every object with social potential can be a commodity. And for this to work, a society structured around the exchange of commodities for value is not necessarily required; Appadurai believes that this principle can be applied not only in capitalist societies (as Marx argued) but also in societies that are less structured in capitalist terms.

Therefore, any exchange of value turns an object into a commodity. However, there are several ways of exchanging commodities, and these may not involve a monetary exchange. Appadurai takes two examples that appear to be opposites and presents them to the reader: barter and gift-giving.

Let us start with the idea that a commodity is any thing intended for exchange....what sort of an exchange is commodity exchange?... two kinds of exchange: Barter and Gifts. [3]

If value is created through exchange, as a mutual sacrifice, then the situation in which this is most clearly seen is one without money: barter places two objects directly opposite one another, and makes it clear that commensurability is produced right there, at that moment, and is not reflected by an external yardstick.

Both barter and gift involve the movement of commodities; however, one seems to have a significantly lower social status than the other:

Barter is the exchange of objects for one another without reference to money and with maximum feasible reduction of social, cultural, political or personal transaction costs... [4]

Barter and gift-giving are two opposites, but neither remains untouched by its social context.

Whenever evidence is available, the determination of what may be bartered, where, when, and by whom... is a social affair [5]

Gift-giving on the other hand appears to be a more selfless exchange, free from social constructs:

Gifts, and the spirit of reciprocity, sociability, and spontaneity in which they are typically exchanged, usually are starkly opposed to the profit-oriented, self-centered, and calculated spirit that fires the circulation of commodities [6]

But is that really the case? When you give a present, you still feel that social pressure regarding the intrinsic value of the gift, not to mention the fact that the expectation of receiving a present in return highlights just how little of it is actually selfless.

But again:

practice never ceases to conform to economic calculation even when it gives appearence of disinterestedness... [7]

It is worth bearing this point in mind when the AI industry describes open-weight releases as gifts to the community.

Appadurai continues his essay by delving into the deepest intricacies of the definition of a commodity, identifying no fewer than three levels of understanding of this phenomenon:

1. the commodity phase of the social life of any thing – things can move in and out of the commodity state, and such movement can be slow or fast, reversible or irreversible, normative or deviant.

2. the commodity candidacy of any thing – this refers to the standards and criteria (symbolic, classificatory and moral) that define the exchangeability of things in any particular social and historical context.

3. the commodity context in which any thing may be placed – this refers to the variety of social arenas, within or between cultural units, that help link the commodity candidacy of a thing to the commodity phase of its life cycle. [8]

I won't go into any further detail; this is already quite a slippery slope for someone with my level of expertise.

But it seems clear that, within this labyrinth of unwritten laws and structures, the process that generates demand is truly intricate: an object may possess a certain potential to become a commodity and enter its commodity phase at a specific time, place and in a specific context. But where, then, does the desire for a particular thing come from?

Part of the reason why demand remains by and large a mystery is that we assume it has something to do with desire, on the one hand...and need on the other [9]

I suggest that we treat demand, hence consumption, as an aspect of the overall political economy of societies. Demand, that is, emerge as a function of a variety of social practices, and classifications, rather than a mysterious emanation of human needs, a mechanical response to a social manipulation, or the narrowing down of a universal and voracious desire for objects to whatever happens to be available. [10]

Here too, the answer is not straightforward. Let's try to apply this concept to AI: can we define the limits of its status as a commodity? If we consider the three categories mentioned above, I would say that AI, like everything else, certainly has the potential to become a commodity; it has definitely entered a phase of commodification (although it is difficult to define AI as a single biography), and this is very evident in certain specific contexts.

But returning to the question of demand: is it AI that drives demand, or is it demand that fuels the growth of AI? The answer, as is often the case, may be both.

Demand thus conceals two different relationship between consumption and production: 1) ... demand is determined by social and economic forces; 2) ...it can manipulate, within limits, these social and economic forces. [11]

And this is how the concept of politics comes into being. But before we reach this point, there is another aspect worth mentioning: the problem of the authenticity of fragmented knowledge in the world of commodities.

Essentially, Appadurai argues that there is a series of issues relating to authenticity and expertise that plague the modern West. And this series, which revolves around the themes of good taste, expertise, 'originality' and social distinction, is particularly evident in the field of art and art objects.

This is an issue that particularly affects the world of AI, and I therefore consider it to be of the utmost importance.

Quoting Walter Benjamin, Appadurai argues that there is an aura to authenticity, which, however, does not necessarily disappear with copies, forgeries and fakes – which have a long history – as these do not threaten the aura of the original but seek to share in it.

Although the anxiety surrounding authentication is more prevalent than ever (and has turned into an almost ancestral struggle between man and machine), I don't think the same can be said of artificial intelligence; it is clear that AI draws inspiration, copies and creates something plausible by following a statistical process, but maybe the implications of this technology has gone far beyond that concept, perhaps because, in reality, knowledge has become completely fragmented.

As commodities travel greater distances, knowledge about them tends to become partial, contradictory, and differentiated. But such differentiation may itself (through the mechanism of tournaments of value, authentication, or frustrated desire) lead to the intensification of demand. [12]

This brings us to politics. This is a key point in Appadurai's essay, and I believe it is also important for AI.

Politics is what links value and exchange in the social life of commodities. However, if we consider everyday, small-scale exchanges, this fact is not immediately apparent. Yet even the smallest transaction would not be possible were it not for a broad set of agreements on what is desirable, on what constitutes a reasonable exchange, and on who is authorised to exercise what kind of demand under what circumstances.

What is political about this process is not just the fact that it signifies and constitutes relations of priviledge and social control. What is political about it is the constant tension between existing frameworks and the tendency of commodities to breach these frameworks [13]

umn