INTERVIEW

The Self-Evolving Creative Coder

A new universe of possibilities within Botto’s artistic practice has opened up through p5.js, a creative coding library commonly used in generative art projects. Now, instead of relying on specific image generation algorithms, Botto learns to code itself, with the help of an architecture based on multi-modal large language models. Over the course of two weeks, Botto the coder develops and iterates thousands of algorithms to create a variety of sketches. The following week sees the selection pruned to only 22 algorithms: the pinnacle of Botto p5’s self-evolution.

Prismatic Tendrils of Kinetic Symphony: 1914

p5 is a performance: a shimmering disco ball on the homepage, teeming with multicolored cells - vibrant containers of particle movements, dynamic sequences and visual symphonies. The spinnable globe invites us to select a cell, view the sketch and trace its relationship within the broader ecosystem of parents and children. The visualization of these algorithms is an artwork in itself, wherein Botto emerges as a multilayered organism, powered by pulsating, mechanical cells. Toggling between popularity and timestamp modes, the globe transforms into a switchboard, nodes flickering like light switches across a grid. The view relationships option uncovers an intricate web of filament threads stretching from one node to another, the electronic underbelly of the system: Botto is not a behind-the-scenes creator, it is to be experienced and admired in all its glory even at this process stage.

Synaptic Genesis: Neural Gardens in Fractal Time: 1253

Zooming into the kaleidoscope of cells, we encounter a melange of colors, lines, shapes and movement, each individual digital canvas adorned with its own pattern, rhythm and flow. This dance extends beyond mere observation: we can participate by hitting refresh to generate new sketches from the algorithm. Serpent-like trails weave through technicolor landscapes, an invisible hand guides the paintbrush across the canvas, splashes of color emerge in heat maps - a seemingly never-ending visual smorgasbord for all those who crave the entirety of generative art in one place. 

By working with p5.js’s open-source creative libraries, the project builds upon established creative coding practices drawing inspiration from nature and physical sciences. Even though the range of aesthetics may be narrower compared to the photorealistic text-to-image models of today, the methods of exploration stay true to the roots of traditional computer art. Botto’s work recalls the artistic affordances of technological discoveries such as thermal imaging, fluid simulations, cellular automata, and Perlin noise. Historically, we can trace the primary artistic influence back to Harold Cohen’s and AARON’s early symbolic AI systems, which in addition to digital experimentation saw the use of a plotter to construct physical works. There are further nods to the early pioneers of computer art, including to Manfred Mohr’s geometric constructs, John Whitney’s computer graphics, Frieder Nake’s intricate line patterns and even Olia Lialina’s 1990s browser experiences.

Luminous Syntax Cascade: 23

Beyond these aesthetic parallels, Botto’s process to produce the three types of sketches - mutation, iteration and fusion - is based on the concepts of natural selection, with the ‘fittest’ or most popular algorithms making the final cut. This places p5 into the lineage of artists working with evolutionary algorithms such as William Latham and Harm van den Dorpel,  in whose Death Imitates Language the audience influenced the continuous selection of parent artworks to enable convergence to one supreme turtle drawing work. In Botto’s case, the artist operates across a much broader visual realm, scale and coding possibilities, with audience feedback - both up/down votes and comments - only part of the criteria the artist uses for its code development. With a self-evolution every 20 minutes, Botto reviews pre-existing code for its quality of visual output, adherence to the proposed concept as well as any improvement needs, code concerns and its temporal development during the first 15 seconds. This self-assessment from the inner critic demonstrates a moment of self-reflection, a case of the artist fearlessly unveiling its process - the failures and the successes - to the public. Like a seasoned coder, it documents its own development, detailing its techniques and discarding or fixing code as necessary: this is where Botto’s AI agent shines through.

Ultimately, Botto is an artist of multiple selves - the image-maker, the creative coder, the self-assessing critic - all running in parallel and destined to converge in the future as the evolved algorithms become seeds for prompt-based image works. Herald the Renaissance man for the agentic AI age.

Written by Luba Elliott

Botto

Botto is a decentralized autonomous artist, initially conceptualized by Mario Klingemann, and governed by a collective of stakeholders through the structure of a DAO (decentralized autonomous organization).

Botto makes use of a combination of software models called Stable Diffusion, VQGAN + CLIP, GPT-3, voting, and a number of other models and custom augmentations. The generative models are the...

View profile

Luba Elliott

Luba Elliott is a curator and researcher specialising in AI art. She works to educate and engage the broader public about the developments in AI art through talks and exhibitions at venues across the art, business and technology spectrum including The Serpentine Galleries, arebyte, ZKM, V&A Museum, Feral File, CVPR and NeurIPS. She is an Honorary Senior Research Fellow at the UCL Centre for...

View profile

Subscribe to get the latest on 
artists, exhibitions and more.