← back

The Machine Seems to Need a Ghost

(but the ghost cannot quite make itself at home in the machine)

Generative artificial intelligence and machine learning are rapidly advancing. Anyone can use image generation tools to create without needing specific technical or artistic skills. The images generated by these tools challenge the notions of work and creator, as if they were algorithmic ready-mades. Like Marcel Duchamp’s urinal, bottle rack or snow shovel, they are products of mechanization and automation (industrial for Duchamp, digital for these new creations) and displayed in an art gallery. The artist does not have to paint, photograph or sculpt; his choices and decisions shape the work. The algorithm draws from a huge database of images that mirror our world without replicating it accurately. The generated images look more and more realistic and close to reality but also act as a distorting mirror, exaggerating all the stereotypes and biases of our visual culture.

We are at a turning point where human production has not yet been contaminated by artificial production. However this will soon change as the tools themselves use their own creations as input. Gradually the feedback loop, an endless cycle where culture ceaseslessley refers to itself, will come to dominate the database, risking getting stuck in nostalgia for the past and trapped in a closed , meta-stable, system. Duchamp’s ghost still haunts us, an unavoidable reference in the history of contemporary art, often quoted, copied or parodied by generations of artists that followed. He became an art cliché despite himself. Duchamp himself described his own art as “meta-irony” to describe his art - a form of critical distance holding its own questioning.

Artificial intelligence raises ethical, artistic and social questions that are only an acceleration of the same questions that have followed the inventions of printing, photography, computer or the internet. The growing automation only makes it harder to escape our current system and the “meta” has become a refuge. This constant self-reference, reflexivity, circularity of our art, our technologies, our culture is becoming a trap where the past’s ghosts still haunt our present thinking.

« The Machine Seems to need a Ghost (but the ghost cannot quite make itself at home in the machine) » is a work in progress currently composed of three typologies that explore all these questions around the keywords of feedback loop, hauntology, meta, etc., as a neural network of linked ideas and images:

→ Variations of Duchamp’s ready-mades, where artificial intelligence industrializes and automates endless mutations of the artist’s most famous works.

→ Images created by feedback loop by feeding recursively the AI tools to create novel-looking visual concepts but which are created only through references to the past.

→ Posters decorated with quotes from artists, philosophers, musicians, critics, who invoke concepts of recursion loops and hauntology.