ABSTRACT
This chapter explores the concept of agency, extending it beyond human actors to include objects and technologies as active participants in networks, drawing on Bruno Latour’s actor-network theory (ANT). I consider large language models (LLMs) as agents, emphasising their use of multi-head attention to process and respond to text. The chapter also discusses the application of agent-based models in urban planning, where computational agents simulate interactions and optimise outcomes. Furthermore, I consider how conversation-centric computing enhances LLM capabilities through task-oriented dialogues, and addresses inter-AI communication, highlighting its potential and limitations for collaborative problem-solving.
