Tacky… cheap… prosaic?

By John Anthony S. Estolloso

At the onset of August, Komisyon sa Wikang Filipino announced this year’s theme for Buwan ng Wika: ‘Wikang Filipino at AI: Kasangkapan sa Pananaliksik at Tulay sa Yaman ng Impormasyon.’ Clearly, the intention was to use artificial intelligence (AI) as a tool for the pursuit of knowledge—until it went overboard with a spate of school-initiated poster-making competitions rendered in AI slop. Inevitably following it was the backlash from the online academic community and the questions about the aesthetics and ethics of artificial intelligence.

To satisfy my curiosity about how large language models (LLMs) and their peripheral apps perceive themselves on these matters, I asked ChatGPT why I should use AI for generating designs. It offered a selection of answers, ranging from the quick generation of ideas to the framing of AI as ‘design partner’, perhaps a euphemism to soften the insipid aesthetics of a bland prompt. Generally, the proposed rationales run on the common themes of instantaneous retrieval of information or assistance in a recommendatory role, always at pace against time and resources.

Aesthetically, AI is bumbling in its attempts to replicate human creativity. At best, what it can do is reconfigure whatever hodgepodge of aesthetic odds and ends fed to it; hence, slop. While multiple apps claim to be a cheap and democratizing alternative to the making of art and design, it precisely reflects those in their results: prosaic to the point of tackiness.

For the argument of the prompt as artistic impetus is contestable. A prompt is a command; it is a relegation of a task or a suggestion to someone or something else that you personally cannot accomplish on your own. A dancer will have the body, the musician his instrument, or the artist his brush, paint, clay, or digital sketchpad as direct mediums of expression; the prompt for a design does nothing of the aforementioned. It has nothing that connects directly the artiste to the execution of the art itself.

While AI provides an easy accessibility to artistic mimesis, it effaces identity from its outputs. Its artificiality diminishes the humanist notion that Art is the most intense mode of individualism. An example: in a recent exhibit at ILOMOCA, a group of artists assembled a collaborative work of art, each painter bringing his style and sensibilities onto the canvas. At the end of the iteration, one can still identify in the frame which artist contributed which aspect of the painting. In all appearances, identity cannot be removed from the artist’s creative process, even if the opus itself is accomplished communally. Something of the artist or auteur becomes embedded in the art. Can we say the same at a glance on a series of AI-generated posters albeit prompted by different sources?

As such, the question of ownership then becomes a subject of contention. Ethically, who claims the work of the machine? Should one declare ownership to an AI-generated document or manuscript fed by raw, inchoate ideas while nary a word was written by yourself? Does the writing and phrasing, literary or formal, on the paper equate to the paltry phrasing of a prompt? Do the melody and harmonies put together by the data bank constitute the work of the human mind?

In a recent Qwertyman article, writer Jose Dalisay, Jr. discussed his grand experiment with AI in a collegiate writing classroom. Instead of banning it outright, he asked students to write compositions as benchmarks; these are then subjected to AI’s perusal and recommendations, and an evaluative comparison ensues on how the app provides alternatives for improvement. Here, the students as writers do not lose grasp of the control on their texts; rather, they are assisted on the conventions and mechanics of the language. In the process, AI remains as a tool.

Quite rightly so. I have no qualms admitting to resorting occasionally to ChatGPT, especially in drafting recurring documents. For instance, rubrics for essay items: why would I waste time filling in twelve to sixteen boxes of descriptors when students would just skim through the key words in the column describing what constitutes the perfect score in a criterion? So I let AI do the menial tasks of academia; but never let it be said that we resort to AI due to an inadequacy of innate creativity or straightforward sloth. We cannot afford the tacky, cheap, and prosaic dependence on something soulless.

* * * * *

A recent conversation among colleagues explored the possibility of using AI in assisting students when translating English texts into Filipino. This reminded me of the Italian adage on translation: traduttore, traditore; translator, traitor—something is always lost in translation. In good faith or otherwise, the translator considers contexts and nuances in rewriting texts and narratives, yet cannot avoid the likely loss of meanings in the process. Now, we propose that a machine will try its hand on this human activity: the irony is not lost when one ponders on what could be more traitorous than removing human sensibility in the transmutation of ideas through language.

Far be it for us to consign AI as artistic pariah and academic anathema, one can always examine its judicious use in art and academe, without compromising the matters through which humanity exudes and enshrines itself. That way, we ensure that the Machine serves Man and not the other way around.

(The writer is a humanities teacher in one of the private schools of the city.)