“85+ ChatGPT Prompts for Graphic Designers,” A.I. Habit
The process of writing is unpleasant and tiresome. Personally, I hate it. The usual position of the body of a writer in front of the computer is unhealthy. It leads to scoliosis and is damaging to the eyes. Writing is a purely manual activity—pressing the letters on the keyboard one after another. One feels oneself in the position of an industrial worker in the nineteenth century. But why are writers still writing? There are different reasons to write but there is a particular aspect of writing that makes it different from other activities. A piece of writing remains stable through time; it promises trans-temporality and even immortality. Of course, that is what connects writing with the visual arts, music, film, and architecture. But all these other practices require a certain kind of organization, a collective effort—and thus money. Writing, on the contrary, is a lonely activity—a last chance for an individual to produce something stable in our unstable world. At least since Derrida it has become untenable to believe that a writer can stabilize their “intention.” Ultimately, we cannot know what an author “wanted to say.” But a piece of writing is a combination of letters—and not of “intentions.” Insofar as this combination of letters is signed, they become “authorized.” And insofar as this combination is literally reproduced—through print, the internet, or otherwise—its author is not dead. That is why authorship survived deconstruction. But will it survive AI?
There is no doubt: the emergence and advancement of AI puts individual authorship in question. The writer—this last artisan amidst the industrialized world—sees their work drowning in an ocean of machine-produced texts. The reader cannot know any more if a particular text is written by a human author or produced by AI. Of course, one can argue that it makes no difference. The reader can also “enjoy” a poem or a novel or become persuaded by a philosophical treatise that is written by AI. And indeed, from the perspective of readership, the difference between human-written and machine-written texts is completely irrelevant. However, does this mean that individual writers are at last relieved from the tedious work of writing?
The answer is: yes—but not quite. Indeed, how does a machine know that it has to produce something? And how does it know what kind of a text to produce? Classical industrial machines produced certain commodities when they were activated—for example, through the push of an electric button—and they produced objects that they were supposed to produce “according to their construction.” One can say that the behavior of industrial machines was instinctive: they were triggered like animals are triggered by external impulses. By contrast, AI must be addressed in writing. Large language models (LLM) are not triggered but prompted. And the art of writing prompts is thus becoming increasingly crucial to the future of human-AI communication. Today, there are courses that teach people how to write prompts. These courses are multiplying. Many of them can be found on the internet. Let us cite a manual from one of those courses:
Prompting for AI is a necessary skill because it allows us to effectively communicate with AI models and get the desired output. By mastering the art of prompting, we can leverage the potential of powerful generative models, such as creating cool stories, amazing images, or other features like text summarizers or automatic video editors while saving time, tokens and in the end: Money.
And further: “A well-crafted prompt ensures that the AI understands the user’s intent, leading to more relevant and accurate responses. Conversely, vague or poorly structured prompts may result in less useful or incorrect outputs.” In other words, the AI-generated text or image is an interpretation of the authorial intent of the prompt. Accordingly, this intent should be formulated clearly so that AI can adequately understand it. In a seemingly paradoxical way, the practice of prompting brings us back to the classical figure of the author—the figure that was put into question by the discourse of deconstruction. Indeed, reading a traditional human-written text, we cannot reconstruct the authorial intention that produced this text. The reader cannot look into the brain of the author to see an initial intention there, and then compare it to the text that this intention generated. However, machine-written texts make this operation possible. One need only look at the original prompt—and then compare it to the interpretation of this prompt produced by the AI. This operation of comparison is equally accessible to the author of the prompt and to readers. But when we speak about the interpretation of the prompt, a question emerges: What does it mean that AI interprets the prompt instead of simply reacting to it?
Hermeneutics, as the analysis of interpretations, is at the heart of the human sciences. The ability to interpret seems to define the difference between humans and things, including machines. Things obey natural laws—for example, the law of gravitation—but humans interpret social laws before following them in this or that way, or not following them at all. That is why humans are historical and things are not. Stones react to the law of gravitation today as they did in the past. However, our contemporary interpretation of the state and its laws went through many radical historical transformations. Thus, if AI operates by interpretation, this means that it is also historical.
Indeed, at every historical moment the functioning of AI as a text- or image-producer is defined by the level of its training, the capability of its technology, but also—and maybe in the first place—by the historically accumulated mass of texts on the basis of which AI operates. As history advances, this mass of texts is changing; some texts are added, some get lost. And the technology of AI is also changing. So if, as a writer, I write a prompt and the AI produces a text or image prompted by this prompt, I can immediately see how my text is understood and interpreted at this particular historical moment—not by a particular individual or group but by the whole civilization in which I live. AI is nothing other than the embodied zeitgeist. And by prompting this zeitgeist-machine, I am able to analyze and diagnose the moment of history to which I am contemporary.
The AI prompting manuals teach the reader to formulate prompts in such a way that their intention is adequately understood and interpreted by the AI. In other words, in a way that ensures that the prompt-writer’s expectations are fully satisfied by the prompted text or image. But then what is the sense in prompting the AI in the first place? Prompting presupposes a reaction that is unexpected, surprising. One prompts AI with the goal of provoking it to write something that the author of the prompt is unable to write. The AI has the ability to process a huge amount of already existing text and images, whereas an individual writer lives in their own textual “bubble.” We have a sense that the bulk of our cultural heritage—textual and visual—escapes our knowledge. Our ability to process the existing textual material is very limited—and this material is perpetually growing in scope. For an individual writer, it is impossible to compete with this growth. So one expects that AI—being able to process much bigger portions of existing information—will respond to a prompt with an answer that reflects an already accumulated mass of writing better than any individual writer could. Insofar as this mass of writing can be seen as an embodied zeitgeist, prompting takes the form of dialogue between an individual author and the zeitgeist.
However, one cannot say that AI manifests something like the vox populi or, as they say now, “the hive mind.” Human life is integrated into cosmic life. Humans have access to the accumulated mass of writing but also to the extralinguistic sphere—to animals, oceans, and stars. Of course, it’s been said time and time again that there is nothing beyond language. But to be able to say this, one has to use a notion of language that is different from the one we are using here: for AI, language is just a mass of written and audio documents—and nothing more. The traditional role of writers is seen precisely in their alleged ability to translate the experiences they have in the extralinguistic cosmic sphere into literary form. To what degree this is possible is an open question. Here it’s enough to say that such a translation automatically cancels itself out at the moment it is attempted: even if a new text pretends to be born out of oceans and stars, it takes its place among other literary products—whereas oceans and stars remain in their usual place outside the literary canon. The degree to which a new literary work breaks with this canon is usually considered the degree of its “authenticity”—of its extraliterary origin. However, in our time one cannot speak any more of a literary canon. The mass of accumulated writing cannot be explored and processed by the human mind. It is experienced as a huge garbage pit into which every new text is thrown as merely an additional piece of garbage. In fact, the permanently growing mass of writing is less accessible to humans than oceans and stars that remain in their places.
People watch TV, visit exhibitions, theaters, cinemas, and bookshops to learn what the zeitgeist—their own contemporaneity—looks like. However, one inevitably has a feeling that one has overlooked vast areas of contemporary life—maybe the most important areas. And this dark area of the zeitgeist infects the souls of “creators,” spectators, and readers with that typically modern nervosity. However, if the accumulated mass of writing and documentation is not accessible to the human mind, it is accessible to AI. Today, prompting seems to be the only way to communicate with this “objectified writing”—this embodied zeitgeist. The prompting manuals that I quoted earlier recommend that readers adapt the style of their prompts to the “clear and distinct” thinking that we tend to associate with “logical” machines. However, as we have seen, AI operates by processing a mass of writing that has been accumulated in a fragmentary and chaotic manner. The logical structure of the texts generated by AI is illusory. To investigate and diagnose the mass of accumulated writing, one has to use not clear and distinct but paradoxical and provocative prompts that put the organizational principles of AI into question and reveal the chaos hidden behind the smooth surface of its results.