Rage against machine learning algorithms

I sit in my uninsulated Paris garret chilled by the freezing winds outside, listening to the oversized rats skittering and squeaking across the floor and in the walls, dipping my quill pen in a bottle of ink that sits beside a funeral taper, my sole source of light, as cheap parchment is spread before me upon my writing desk.

And suddenly the ghosts of Geoffrey Chaucer, William Shakespeare, Samuel Butler, Lord Byron, Walt Whitman, Emily Dickinson, TS Eliot, Allen Ginsberg, and Sylvia Plath materialise before me, their ectoplasmic slime dripping on the floor and the ghoulish green light of their spectres emanating from no earthly source. 

The years pass by but they never leave. They look over my shoulder and make comments as I read, write, look at something, hear anything, smell or touch or taste. I am not the only one haunted by them. Their poems are incantations that bring forth their shades. 

And they demand from me and everyone else they haunt some kind of accountability of the world.

If people prefer AI-written poetry to human-written poetry, they are wrong. 

If people do not have the time or interest in human-written poetry, that is their fault. 

Mass appeal is neither an objective nor a reasonable measure of quality or aesthetic beauty. 

What the people believe to be beautiful is not necessarily beautiful simply because they believe it to be so.

In the field of logic, argumentum ad populum, an argument that appeals to the people or to popular belief, has long been widely recognised as a core logical fallacy, that is, as an unsound or inadequate basis for proving anything true. 

A tech billionaire's courtier or fanatic may conceivably argue that AI-generated poetry is indeed better, more beautiful, than human-generated poetry simply because the people believe it to be so, but such an argument is without basis and proves nothing.

When we have become accustomed enough to our righteous indignation to be able to analyse rationally once again, we can also blame society – a deficient public education system that, on the one hand, is underfunded. On the other hand, it devalues the humanities, sometimes to the point of not even requiring students to read an entire book, due to its unwarranted focus on standardised tests and preparation for a corporate career – a deficient neo-liberal university (that in the US is increasingly expensive and unaffordable).

It devalues the humanities and that cares more about its own profits than it cares about producing critical thinkers – a deficient neo-liberal culture obsessed with celebrity and profit, replete with film production companies and publishing houses that churn out garbage meticulously designed to not push the limits.

Not only poets, but all creative writers ought to strive for maximum complexity and opacity. They ought to strive to never be straightforward, accessible, easy to understand or interpret. The creative writer should not try to communicate. At the heart of creative writing is the incommunicable. 

The creative writer should not want to be understood. To be understood is a mark of deficiency – a deficiency of imagination, a deficiency of ambition, a deficiency of passion. 

AI, too, is deficient in all of those things, since it is, after all, a series of mathematical calculations and not a human being. 

And despite what science fiction authors and cognitive scientists in the West have been telling us since the beginning of the Cold War, human beings are not reducible to a series of mathematical calculations. 

In any case, no one has yet succeeded in reconstructing a human being from computer algorithms, and such a prospect, if indeed even possible, is still a long, long way off.

Arnold Schwarzenegger in the "Terminator" films is a human pretending to be a robot, and not an actual robot. The same goes, mutatis mutandis, for the Westworld series. The illusion may appear convincing, but ultimately it is merely an illusion, and it is a deliberate one crafted by human beings. Human beings provide the faces, the voices, and the thoughts for your favourite fictional robots.

Creative writing is not about communicating an image, a theme, or even an emotion. 

Creative writing is not about atmosphere. 

Creative writing is about thought, and more particularly, human thought, which humans themselves don't completely understand, whether one reflects upon one's own thoughts or another's thoughts.

Human thought involves not only language and not only consciousness, but also the unconscious, and the unconscious is something that even the psychoanalyst does not fully comprehend. That's why there are so many competing schools of psychoanalysis – Freudians, Jungians, Kleinians, Lacanians, etc. 

The unconscious is only knowable through consciousness, so we cannot know about all those regions of the unconscious which the unconscious has not revealed to consciousness. 

While animals lack language (you cannot teach them English, Latin, Bangla, or your local village dialect in any meaningful sense) and, arguably, an unconscious (which we may infer since they don't have any cultural taboos), computer algorithms have none of the above – neither language, nor consciousness, nor an unconscious.

Large Language Models (LLM) string words together based on a series of mathematical calculations, heavily involving linear algebra, through which it deduces which word is the most statistically likely as the next word. 

That is a statistical approximation of language, but it is not yet language. Human beings, after all, don't string words together in this way, performing these same equations. No adequate mathematical models of language, consciousness, or the unconscious yet exist.

Relying on LLMs to generate writing deprives the human writer of the joy of creation, the pleasures of the imagination. The negative pleasure is nothingness, as opposed to the positive pleasures of being: the pleasure to be nowhere, to be nobody, to do nothing, to not be productive, to not communicate anything. 

The pleasure of the imagination, which is the pleasure of creativity, is also like the pleasure of dreaming, the pleasure of thoughts that seem to come out of nowhere, the pleasure of being in touch with the unconscious.

The poet's struggle against LLMs is both a struggle against tech-bro capitalists, a small but mostly unaccountable group of people who control a vast global communication infrastructure and are primarily driven primarily by their need for profit at any cost, and a struggle for one's own consciousness. 

The poet is faced with a choice between relying on LLMs to produce texts – thereby reducing the pleasure available to consciousness and replacing thoughts with statistical approximations of language ultimately overseen by billionaire tech-bros with dubious ethics and politics – and relying on consciousness and the unconscious to accomplish the difficult task of thinking at the price of being increasingly alienated from other people who will neither understand nor like them.