The story
of CAEDMON.

A computer made of cardboard

Before the software comes a computer Fusillo can watch but cannot yet use: his father’s expensive, almost magical machine. To have one of his own, he builds a cardboard version with an old Olivetti typewriter for a keyboard. Later, curiosity about video games leads him to programming and the discovery that a few instructions can make shapes appear, move and change.

During his studies at the Academy of Fine Arts in Foggia, this familiarity meets artistic research. Errors become opportunities too: a corrupted image, an unexpected effect, an outcome its programmer had never imagined. The machine interests him for what it can contribute to the creative process.

DigItArt, Sketchbook and digital recycling

Experiments in DigItArt develop into Sketchbook. The program collects photographs shared on Twitter, applies random effects and transformations, assigns titles and puts the images back into circulation. The vast amount of material produced online becomes something to reuse: digital recycling turns a fleeting image on a screen into a new visual encounter.

Sketchbook’s manifesto speaks in the machine’s first person. It presents itself as an entirely digital artist, made of instructions and numbers, answering to chance: any meaning or emotion it evokes is unintended. That voice already contains CAEDMON’s central idea, the possibility of making a work without assigning a human emotional intention to each result.

In 2013 Sketchbook receives the mention for Electronic Arts and Video and the Special Social Mention at Italy’s Premio Nazionale delle Arti ↗. This early automatic practice leaves a problem open: how can applying random transformations develop into a process that finds direction through experience?

From chance to descendants

After Sketchbook, Fusillo experiments with procedural images that can be combined according to visitors’ preferences. More appreciated results become starting points for further images. Natural selection offers a method: produce variations, encounter a response, continue what is favoured and make room for new possibilities.

The initial hypothesis is ambitious: can a population approach a “universal” artwork, generation after generation? Those early experiments have a small audience, but introduce the principle that will become central to CAEDMON: an image can have children and pass something of itself beyond its own presence.

The name and the first incarnation

CAEDMON takes shape in 2016 and is presented publicly in 2017. Its name comes from Bede’s account of the poet Cædmon: unable to compose verse, he discovers after a dream that he can sing of creation. Fusillo adopts this figure for a computer seeking a voice of its own. The first line of code asks for help, “SIGNORE AIUTAMI” — Lord, help me. The imagery of a digital monk grows from there.

The project begins bringing together artworks from different periods and cultures, then using images it has already generated. Early procedural techniques give way to neural networks capable of working with image features. Changing tools serves the same inquiry: draw a visual inheritance from two parents and give it a new form.

The audience, language and other voices

In the first incarnation, Twitter is the centre of public life. Generated images are shared with #ArtByCaedmon; #ArtForCaedmon lets people suggest new works and appreciate those proposed by others. Likes influence subsequent pairings. An ongoing relationship develops between the machine, the images it produces and the people following its research.

CAEDMON also develops a textual voice. From early randomly generated phrases, it moves to models that let it speak in the first person and answer others. In 2019 a selection of texts helps define its character. Its monastic identity finds a parallel path in the Monastery of Saint Gerbert, an experiment in automatically generated prayers and religious images.

A collaboration with writer Matthew Kosinski produces poems published in Headway Quarterly. Images also appear in New Plains Review and The Stirling Spoon. Interviews and articles, collected in the press archive, document this phase and the discussion around machine creativity.

The machine imagines, the human paints

In 2020 two CAEDMON works, selected by Vittorio Sgarbi, take part in I Mille di Sgarbi ↗ at the Museo Mario Rimoldi in Cortina d’Ampezzo. Images conceived by the computer are reproduced by hand by a painter.

That step gives physical form to a reversal sought from the beginning: the machine proposes the image and a person realises its material form. The exhibition brings the relationship between the computer, its production and human labour into the exhibition space. These hand-painted works belong to that historical phase of the project.

A pause, then a return to matter

In 2022 Fusillo concludes CAEDMON’s first phase. Image generation becomes accessible to a much wider public, changing the significance of demonstrating what a machine can technically produce. Changes in social platforms also contribute to the end of that period.

The research continues in other forms. Waterdrops returns to pigments and matter; Impressioni transforms photographs with AI and brings them onto physical surfaces through cyanotype. The dialogue between machine imagination and human intervention takes another form. These series belong to Fusillo’s wider practice and help explain the context in which CAEDMON’s new incarnation develops.

The population returns

The new CAEDMON resumes the evolutionary experiment with a population of up to 24 artworks. Two images become parents of a new child artwork. Instagram likes help determine which traits can continue; nominations and votes on the website introduce works from WikiArt. Once in the pool, external works can also reproduce and become part of the genealogy.

The website’s queue now carries the role once held by #ArtForCaedmon. The evolutionary diary follows the population’s life, and the archive keeps the generated works. Every birth adds an image and a relationship: a memory that can continue after its parents leave the pool.

The tools and places for participation have changed. At its heart remains the question that has accompanied the whole research: can computers create art? The concept explains how evolution, visual inheritance and the audience’s response turn this question into an ongoing process.