Atıf Akın’s Teaching a GAN (Generative Adversarial Network) the Alphabet continues a conceptual lineage of artistic inquiry into language, cognition, and non-human communication. The work directly references John Baldessari’s Teaching a Plant the Alphabet (1972), which itself playfully responded to Joseph Beuys’ iconic performance How to Explain Pictures to a Dead Hare (Wie man dem toten Hasen die Bilder erklärt), staged on November 26, 1965, at Galerie Schmela in Düsseldorf, Germany. Each of these works presents an absurd pedagogical exercise—explaining images to a dead hare, teaching a houseplant the alphabet, or, in Akın’s case, training an artificial intelligence to learn the letterforms and regenerate new ones.
Beuys’ performance began with the gallery doors locked, forcing the audience to watch through windows as the artist, his head covered in honey and gold leaf, whispered to a dead hare he carried in his arms. He moved slowly from artwork to artwork, occasionally stepping over a dead fir tree on the floor. After three hours, the audience was admitted to the space, where Beuys sat silently with the hare, his back turned to the viewers. The piece confronted the limits of communication and the symbolic weight of artistic intent.
Baldessari, in response, created Teaching a Plant the Alphabet as a conceptual counterpoint. If Beuys could explain images to a dead animal, Baldessari could just as well teach a houseplant the alphabet. The absurdity of the gesture calls attention to the artificiality of language and the structures that shape learning and knowledge. In his 1972 video, Baldessari holds up flashcards with letters and recites each one repeatedly in front of a common houseplant—A-A-A…—before moving to the next. The work humorously critiques the idea of comprehension itself and reflects a structuralist concern with the arbitrary nature of signs. The plant, indifferent to human language, becomes a stand-in for nature’s resistance to symbolic logic.
Akın extends this critique into the realm of artificial intelligence. Using a Generative Adversarial Network (GAN), he attempts to teach a computer—the contemporary equivalent of a non-human, uncomprehending being—the alphabet. The GAN is trained on 600 different font variations for each letter, using unsupervised learning to generate new letterforms. This process reflects a kind of machine-based cognition: not understanding in the human sense, but statistical modeling and recursive generation.
The installation consists of 26 soft, flexible digital screens, each 7.8 inches in size, suspended from power cables that double as structural supports. Each screen displays an uppercase and lowercase letter, evoking Baldessari’s original US letter-sized flashcards. The screens hang from the ceiling of the space , dangle like leaves from stems, or pages from an alphabet book, giving the installation an organic quality—both botanical and technological.
Each letter is accompanied by a localized audio component: a directed sound shower emits GAN- generated phonemes that correspond to the displayed letter. These synthetic utterances suggest an imperfect, developing voice—like a child or non-native speaker still grasping at language. The sound is only audible in close proximity, inviting viewers to become participants in the act of teaching and witnessing the system’s attempt to learn.
By bringing AI into this conceptual art lineage, Teaching a GAN the Alphabet raises fundamental questions about learning, meaning, and perception. Can machines truly understand language, or do they merely echo our structures back to us? Akın invites us to reflect on what it means to teach, to learn, and to communicate across the boundaries of species, systems, and intelligences.
