From Mental Models to Algorithmic Imaginaries to Co-Constructive Mental Models
From Mental Models to Algorithmic Imaginaries to Co-Constructive Mental Models
Abstract
Not only in the course of technological advancements in the fields of machine learning and artificial neural networks in recent years, but also due to an increasingly widespread public debate regarding the regulation and transparency of socalled artificial intelligence, as recently demonstrated by the debates around generative AIs such as ChatGPT or Dall-E, there are increasing demands for the explainability of AI. But almost all work in the field of Explainable AI (XAI) is aimed solely at the perspective of researchers and developers and their intuitions of what constitutes a good explanation. This problematic perspective is also reflected in theoretical concepts important for the development of AI systems, such as the concept of mental models originating from cognitive science. As part of a coconstructive XAI research, and based on two central texts by Kenneth Craik and Donald Norman, this paper aims to argue for a reconceptualization of such models, widely received in computer science and human-computer interaction, using the media cultural studies concept of algorithmic imaginaries, and to point out future lines of research.
Citation
Copy Citation
BibTex
Except where otherwise noted, this item's license is described as:
Creative Commons Attribution Share Alike 4.0 Generic
Creative Commons Attribution Share Alike 4.0 Generic
Keywords:
Machine LearningArtificial Neural NetworksArtificial IntelligenceDebates