Chapter

The Moon’s Reflection on the Water and the Flower’s Reflection in a Mirror

Latent Time, the Black Box, and the Interface in Chinese AI Art

EXCERPT

The recent rise of the foundation model in machine learning, with its exquisite yet immensely black-boxed mode of interaction, necessitates a shift of focus in our discussions of AI art. Until now, AI art has typically been approached as an amalgam of technical skill, artistic endeavour, and critique that delves into questions of data politics, labour, bias, and values through an exploration of both the material and cognitive infrastructures of AI. But the inversion of the power dynamic between the artist and the algorithm that comes with these newer models (locked out of the interior workings of AI software, artists nowadays engineer prompts rather than customising or fine-tuning training data) and changes in machine learning architecture (such as the elimination of supervised learning and data labelling) are challenging these established approaches, especially those that rely on having access to the technical aspects of the systems in question.

On the one hand, the presence of AI in our lives is now no longer limited to its use in the computer industry or to specific contexts and tasks—it has become diffuse and atmospheric, operating not as a solidified set of tools but as a series of heterogeneous, fragmented processes through which divergent perceptual and phenomenological experiences are triggered. On the other, artistic processes in general are not readily reducible to pre-given parameters or graspable as mere functions of models, rather, they occupy a vibratory plane at the intersection of technical, social, and cultural influences. Therefore, moving beyond the prioritisation of the technical in our approaches to AI may allow some alternative dimensions, even those at the edge of attention, to surface, creating centrifugal tendencies away from the dominant narrative of artificial intelligence in which the technology simply follows a global trajectory of development—from the pre-GAN era to the GAN era to the era of the foundation model, and so on. Having an awareness of the multiplicity of systems of meaning surrounding AI and how they derive from local historical, cultural, and affective particularities—as opposed to assuming the existence of a single universal narrative—will come to play an important role in our understanding of the future (or non-future) of technology on our planet.