Stochastic Remembering and Distributed Mnemonic Agency

Abstract This paper introduces the concept of stochastic remembering and uses two prompt engineering techniques to critically examine the text generated by ai chatbots. These techniques – step-by-step prompting and chain of thought reasoning – are then experimentally applied to understand how ChatGPT, the most commonly used ai chatbot, shapes how we remember historical activists. This experiment suggests that hegemonic forms of memory influence the data on which these chatbots are trained and underlines how stochastic patterns affect how humans and ai systems collectively remember the past. Humans and ai systems prompt each other to remember. In conclusion, the paper argues that ai chatbots are a new kind of mnemonic actor that, in interaction with users, renders a probabilistic past. Methodologically, the paper introduces, in an explorative way, an experimental method that can reveal the dynamics of stochastic remembering.