This page requires Javascript. Please enable it to view the website.

The AI Picbreeder Experiment

Best Paper nominee, GECCO 2026.

Sam Earle*<br>NYU

Kai Arulkumaran<br>Sakana AI

Andrew Dai*<br>Independent

Akarsh Kumar*<br>MIT

Julian Togelius<br>NYU

Sebastian Risi<br>Sakana AI

Jul 6

2026

Paper

Code

Data

* Work done during an internship or residency at Sakana AI.

slop, fine art is punctuated by new AI-driven forms, and every day we encounter of all manner of uncanny objects falling somewhere between.<br>--><br>Can AI agents be creative? It's a question on the lips and fingertips of many.

Sure, AI is the new medium, but could it ever be the one making the message?

We ask whether the agents we use today as tools might be re-purposed to spontaneously speak among themselves.

The end result would be a model organism of cultural production.

What should we ask of the agents to make this real?<br>The tension is that we cannot ask for anything at all—it has to be left up to them—yet modern agents demand our asking by design.<br>They are trained, evaluated, and orchestrated as goal-following entities.

But creativity elides planning, and the end products of cultural processes are not conceived of at the outset but forged in serendipity.<br>This is true not only of the arts, but of math and science as well, where the discovery of new questions is at least as important as their resolution.

To this end, we envision agents capable of intentional aimless wandering—agents that can surprise themselves, and in response throw their plans and preconceptions out the window in pursuit of the unexpected.

(Not merely a question of the arts, creativity is a pre-requisite for any effort to seriously automate math or science, which involve not only solutions but the posing of new questions, which themselves follow a more subliminal set of incentives.<br>Processes of collective scientific pursuit, in other words—far from mere optimization processes over the objective of explaining experience in the world—are themselves creative cultural organisms.)

-->

--><br>To study whether AI systems might have this capacity, we turn to a minimal substrate where we know it to be possible—even necessary. Picbreeder was a website which gained a modest cult following in the 2010s. Here, you might imagine us trying to re-animate it with something like the mechanical ghosts of its past users: replacing humans with large models trained on their collective output, and asking if these models are capable of the same level of creative discovery as we were.

At picbreeder.org , users participated in a process of generating collaborative art by interactively evolving images. The images began abstract, but over multiple sessions with different users, genealogies complexified and familiar forms emerged, often by surprise. (Image retrieved via the Internet Archive.) (Click to access an homage to the original site, populated with aggregate results over multiple replays of AI-driven reproductions of the collective human event.)

Before frontier models allowed us to speak into existence a vast distribution of possible images, Picbreeder had users breed images by representing them indirectly.<br>Each image in Picbreeder is encoded as a Compositional Pattern-Producing Network (CPPN), an initially small, protean neural network that takes as input some coordinates in pixel space, and outputs that pixel's color. Because they can take arbitrary continuous coordinates as input, these networks effectively encode infinite-resolution images. Elsewhere, they've been used to represent videos, terrain in game-like environments, and even the weights of larger downstream neural networks with more regular topologies.<br>You can see how coordinate values are turned into color pixels in the animation below:

(x,y,r) -> CPPN -> output pixel<br>* (B) an interactive viewer: pick a human/AI genome and watch it render,<br>* hover a pixel to see its (x,y,r) coordinate map to a colour.<br>* Both reuse the in-browser CPPN engine (breed/cppn.js, pixel-identical to<br>* the paper) and a small curated bundle (breed/data/cppn_explainer.json,<br>* built by tools/build_cppn_explainer.py).<br>* =================================================================== -->

&rarr;

&rarr;

&rarr;

&rarr;

Compositional Pattern-Producing Networks (CPPNs) At every pixel,<br>x,<br>y coordinates, and radius from center r are fed into the network, which returns hue, saturation and value (HSV), i.e. a color. In this way, each CPPN encodes an image of arbitrary resolution. Click the input grid to pin the probe in place; press 'r' to reset.

In practice, we pass all the input coordinates through the network at once.<br>Each node in the network has some grid-shaped activation we can render as a greyscale image, and at the network's output, three such outputs are combined to produce a single color image.

In the tool below, you can see how changing the weights or activation functions of various nodes and edges changes the intermediary activations and final output...