Abstract:

The present study employs a collaborative autoethnographic approach, integrating the insights of a postdoctoral researcher lacking expertise in technology with those of a professor of interactive technology. Together, they developed a methodological approach to programming AI using natural language and transmedia interactions, drawing on the ethos of play as described by Johan Huizinga Their collaboration was sparked by a significant challenge: the absence of a physical robot as a research tool for exploring human-AI interaction in educational contexts. In response, the researchers created a digital assistant specifically designed for kindergarten education, leveraging ChatGPT as a foundational tool in its development. Immersed in this process, the researchers developed more than an educational assistant for kindergarten: they managed to allow ChatGPT to shape itself by providing transmedia inputs. If you consider the analogy of how we interact with AI today, it is like sending a text request to a person locked in a dark and soundless room, who possesses a certain amount of information, has time to spare and can use logic and reasoning to respond. By enabling the AI algorithm to access multiple environmental inputs (e.g., the voices of those interacting in the room, photographs of the room, and the wider context of the task) and instructing it to interact in order to ascertain more information about those involved in the process (e.g., their names, preferences, research interests, etc.), a methodology was developed through a transmedia lens. Two key aspects are highlighted by this methodology: (1) the value of interdisciplinary collaboration in enhancing our understanding of AI's nature, and (2) the conceptualization of AI as a continuously evolving transmedia narrative, shaped by multisensory stimuli—such as images and sounds—and ongoing human interaction.

Keywords:

Programming, AI literacy, play, collaborative autoethnography, creativity, education.