Cyborg-Phagia
II ·· Anthropophagic Manifesto (Phagia)
One of the simplest ways to imagine how large language models, or LLMs, operate is to think of them as a digestive system: the LLM receives our words, deciphers them, processes them, and returns an output based on the text we previously gave it.
This metaphor brought back to me the premises that Oswald de Andrade formulated through the Anthropophagic Manifesto, published in 1928. Oswald, a poet, essayist, playwright, novelist, journalist, and theater critic, was a key intellectual and theorist of Brazilian modernism, as well as one of the organizers of the 1922 Modern Art Week. The manifesto originally appeared in the Revista de Antropofagia, and became one of the most cited texts in Brazil's twentieth-century cultural imaginary.
The text is a contestatory declaration against an artistic elite positioned at the heart of Brazilian artistic and cultural production, whose aesthetic and stylistic values were deeply based on the idealization of European beauty canons. Its call was to rebel against that external aesthetic authority through a cannibal practice.
The most representative phrase of the text is a reconfiguration, a recycling, of Shakespeare's classic phrase "to be or not to be." De Andrade takes it, processes it, deciphers it, and expels it back transformed into a Brazilian linguistic icon: "Tupi or not Tupi, that is the question."
In that gesture, he transforms the European-Anglo "being," the "to be," into "Tupi": the situated subject of Amazonian Brazil, the Indigenous being. In this way, the Anthropophagic Manifesto, more than an aesthetic manifesto, becomes a decolonial metabolic machine that offers an alternative to the mere rejection of the European canon. Instead, it chooses to consume it, metabolize it, and transform it in favor of the creation of a new symbol: an object of local Brazilian expression.
The comparison I propose between De Andrade's cannibal practice and the operation of language models carries a risk: depoliticizing the cannibal gesture, anthropophagy. Language models have no bodily experience and are not subjected to the needs and situations proper to our human nature; therefore, the analogy could slide toward a kind of bulimia, where the automatic repetition of consumption turns the gesture into empty repetition, stripped of the most important action in anthropophagic practice: symbolic appropriation.
But this is resolved the moment we stop thinking of language models as a complete digestive system and instead position them within a symbolic chain of meaning production, as one of the elements that could help us execute our anthropophagic practice as situated subjects.
And this subject is not just any user.
It is here that human existence itself, lived experience, rises as a political element. Because it is not the LLM that defines the meaning of what it produces: it is we, as situated users, aware of our perspectives, experiences, preferences, opinions, and reasoning, who decide the validity of the outputs it gives us.
Consequently, this decision is not neutral. That is what makes human existence political in itself, since a situated user is not a passive consumer or an admirer of the model's statistics. Rather, it is a user who enters the platform to stress the LLM, to demand positionality, to demand that their own imaginary direct the metabolization to be executed. This user does not come to accommodate themselves to the language model; they demand the necessary calibration according to their own experience.
Therefore, it is not enough to use these technologies passively. What is required is to use them from subalternity in order to prevent the center from continuing to replicate itself.
Using artificial intelligence as a situated user is a gesture that twists the Anglocentrism of models created by large corporations and forces them to work with us as collaborators in an anthropophagic practice.