Abstract:

Previous research in the field of lifelong learning has already established that learning for life occurs when the individual interacts meaningfully with the wider social environment. Conversation Theory (CT), developed by cybernetician and educational psychologist Gordon Pask in the 1970s, has been recently articulated as a valuable perspective to investigate the performance of social generative artificial intelligence (GenAI), and to leverage the affordances of GenAI in ethical ways for learning. Conversational theory basically sets up a system within which to view learning, drawing on concepts from the fields of artificial intelligence and computer-aided instruction. Central to the theory is the idea that it is no longer possible to make a clear distinction between apprentice and expert in describing two participants in a conversation which leads to learning. For Pask, the brain of the person who is learning can operate in two distinct modes which can be viewed as ‘expert’ (directing attention to what needs to be done) and ‘apprentice’ (assimilating the subject matter). In CT, learning requires at least two participants who construct understanding out of a heuristic process called teachback, which links elicitation to the requirements of representing knowledge. For that purpose, CT provides criteria for someone to say that he/she understands an ordinary expert’s explanation. To put in other words, Conversation Theory lays down a foundation for eliciting knowledge from an ordinary person as if he/she was playing the role of an expert. When conversations about a subject matter are systematically replicated in a social networking space, not only individual learning takes place, as knowledge becomes stabilized in an entailment mesh of concepts with different perspectives. In this paper, we focus on digital stories in which characters in a conversation try to seek shared understanding. We look at stories scripted and plotted from three broad positions: one in which the main character does not seem to care about reaching shared understanding; a second one in which the character does not find easy to follow the process; and third, one in which the process created for setting shared understanding flows smoothly. Our approach is exemplified by a series of Do-It-Yourself (DIY) short animations in the domain of agroecology. Stories are an excellent opportunity for transcending the barriers reality places on our ability to inhabit the lived perspective of other agents’ experience, whether humans or machines. Although there are undeniable conceptual differences between people and artificial agents, we argue that the inference model of conversation represented in the animations offers a framework for taking the relationship between people and GenAI as inherently social. We finish up by addressing questions deserving further investigation.

Keywords:

Conversation Theory, learning in a social networking space, generative AI, DIY animations, agroecology.