It is a well known phenomenon that the collective behaviour of ants is achieved through the diffusion of pheromones secreted by individuals. The authors applied this mechanism to interactive visual stage effects for a contemporary ballet choreographed by Jiří Kylián at the Netherlands Dance Theatre in 2008. The dance performance entitled “Gods and Dogs” has been performed in various European dance companies. In 2023, the software underwent a complete revision to adapt to the latest computing and peripheral equipment. By utilising the division of tasks between multi-core CPUs and GPUs, the software can now respond with shorter latency and generate visual output at higher frame rates in high resolution. Parameter tuning is also crucial in matching the visuals to the designer’s preferences. The experimental results regarding the tendency and sensitivity of certain parameters for diffusion, as well as the balance factor between attraction and repulsion, can aid the design process.

This article introduces an experimental artwork that employs a reinforcement learning algorithm as core element for an interactive and aesthetic experience. The learning algorithm involves a simple navigation task for a single agent. The agent’s learning process is made perceivable to visitors by animating and visualizing a massive particle system on which the agent’s memory acts as force field. Through interaction, visitors can either facilitate or hamper the agent’s learning process. The goal of the artwork is to convey in a playful manner the increasingly intertwined coexistence between humans and artificially intelligent entities.