The current artificial intelligence paradigm, where exponential scaling of data and computation leads to an increase in functionality and application, requires exponential demand in energy usage, data storage and raw materials for computing components. We compare this dynamic of resource dependence and depletion of AI systems to population dynamics of a snail, the symbol of the degrowth movement. Juxtaposing both phenomena as autopoietic systems within a structural coupling with their environment we identify the difference between sustainable and unsustainable coupling, meaning the ability to sustain itself over time. Due to an absence of negative feedback loops for AI systems, with resources ultimately limited, we identify the state of current AI systems and resources as unsustainably coupled. As AI systems are currently in a process of homogenization in form and function, we call for exploring alternative ways of being for AI systems, for example inspired by the sustainable dynamics of snails within their ecosystem.

Chemical Organization Theory (COT) is a framework to study the relation between structure and stability in reaction networks. It combines structural and stoichiometric conditions underlying self-production, and identifies a class of subnetworks on each reaction network, so-called organizations, that can be mapped to the possible limit sets that exhibit dynamical stability. So far, organizations have been applied as a model for the emergence of autopoietic systems in biochemistry and other areas, but there is currently no systematic study on how organizations can become more complex. Here we formalize the dynamics of transitions between organizations as a Markov process where nodes of the process graph are organizations and transition probabilities reflect the probability of transition from one organization to another. Hence, the structural evolution of a reaction network is seen as a random walk in the graph of organizations. We introduce the terms local and global resilience to describe organizations that have a tendency to resist perturbations or are more likely to be visited on a random walk. This allows a more sophisticated investigation of the influence of structural properties on the evolution of chemical organisations.