From mycelia to machines: categorizing AI systems using fungi-inspired analogies

What is it that AI systems can do? To what extent can they act like us? To answer such questions, we usually adopt an approach in which we use conceptual frameworks to study AI systems. As I show in this paper, this method contains an implicit assumption which becomes problematic as soon as the approach is adopted to study AI systems. This assumption is that it is clear what the term ‘AI-system’ actually refers to. I show that this is not the case and that this prevents us from having meaningful and productive discussions about what AI systems can do or what their impact on our society might be. What we need, then, are accessible conceptual tools that allow technical and non-technical experts alike to specify what kind of AI system they want to study with the help of their frameworks. In this paper, I show that with the help of several fungi-inspired analogies we can obtain such tools and I demonstrate how they can be used to categorize different AI systems without requiring detailed technological knowledge. Taken together, these tools provide a framework allowing for inclusive and meaningful discussions about the societal impacts of AI systems.