No document available.
Abstract :
[en] In his latest work, Der Konjunktiv der Bilder. Meine virtuelle Kamera (K.I.) (Spektor Books, 2024), the German filmmaker, philosopher, and writer Alexander Kluge turns his attention to AI-generated images. Throughout his career spanning more than 60 years, Kluge has consistently engaged with innovative devices. It is therefore unsurprising that he has taken an interest in what he calls a new “tool.” The structure of his work is equally predictable, as it aligns with his numerous essays, experimental montages, and collaborative books created with visual artists. Der Konjunktiv der Bilder indeed presents a heuristic montage of visual pieces and textual fragments whose discursive registers oscillate between poetic evocation, philosophical quotations, and anecdotal narratives. What is far more surprising, however, is the perspective on AI that gradually emerges from this complex interplay. After dedicating a few pages to his theory of cinema - rooted in his early Ulm writings and defined by a distinctive concept of a “principle of cinema” based on associative and constellatory work - Kluge first invokes the photograph of a cardboard reconstitution of a highly publicized place (Backyard by Thomas Demand) to introduce the free associative processes he discovers in his initial experiments with Stable Diffusion. Suggesting an urgent need to explore the AI device’s potential while it still produces errors, Kluge then succinctly claims that what he terms the “virtual camera” is freer than the traditional film camera. He illustrates this assertion by presenting a series of dysfunctional images generated through his minimalist prompts, which, in his view, demonstrate that in its unpolished state, Stable Diffusion can produce genuine heuristic montages. These montages, he argues, surpass the capacities of traditional cinema and offer fleeting glimpses into his “principle of cinema.” Rather than criticizing a device lacking human utopia, wishes, and invention (all possible variations of what he calls the “Konjunktiv,” the subjunctive, of pictures), Kluge surprises his reader by proposing that the AI device not only unveils very rich expressions of the subjunctive through visual associations but, more provocatively, reveals a new version of the Benjaminian aura. This presentation aims to unfold this enthusiastic yet speculative perspective in three stages. First, it briefly outlines the foundations of Kluge’s distinctive theory of cinema, focusing on key concepts such as associations, constellations, and wishes. Second, it comments upon some of the results of Kluge’s experiments with Stable Diffusion. Finally, in its third part, I will, following Kluge’s lead, it tries to elucidate how the concepts of aura and constellation might indeed find a fertile actualization and transformation when confronted with generative text- and image-based AI. This not only renews our perspective on such technologies but also reopens the question of the contemporary relevance of Benjaminian concepts in a visual environment increasingly populated by AI-generated images.
