A widespread need to explain the behavior and outcomes of AI-based systems has emerged, due to their ubiquitous presence. Thus, providing renewed momentum to the relatively new research area of eXplainable AI (XAI). Nowadays, the importance of XAI lies in the fact that the increasing control transference to this kind of system for decision making -or, at least, its use for assisting executive stakeholders- already affects many sensitive realms (as in Politics, Social Sciences, or Law). The decision-making power handover to (...) opaque AI systems makes mandatory explaining those, primarily in application scenarios where the stakeholders are unaware of both the high technology applied and the basic principles governing the technological solutions. The issue should not be reduced to a merely technical problem; the explainer would be compelled to transmit richer knowledge about the system (including its role within the informational ecosystem where he/she works). To achieve such an aim, the explainer could exploit, if necessary, practices from other scientific and humanistic areas. The first aim of the paper is to emphasize and justify the need for a multidisciplinary approach that is beneficiated from part of the scientific and philosophical corpus on Explaining, underscoring the particular nuances of the issue within the field of Data Science. The second objective is to develop some arguments justifying the authors’ bet by a more relevant role of ideas inspired by, on the one hand, formal techniques from Knowledge Representation and Reasoning, and on the other hand, the modeling of human reasoning when facing the explanation. This way, explaining modeling practices would seek a sound balance between the pure technical justification and the explainer-explainee agreement. (shrink)

To explain why, in scientific problem solving, a diagram can be “worth ten thousand words,” Jill Larkin and Herbert Simon (1987) relied on a computer model: two representations can be “informationally” equivalent but differ “computationally,” just as the same data can be encoded in a computer in multiple ways, more or less suited to different kinds of processing. The roots of this proposal lay in cognitive psychology, more precisely in the “imagery debate” of the 1970s on whether there are image-like (...) mental representations. Simon (1972, 1978) hoped to solve this debate by thoroughly reducing the differences between forms of mental representations (e.g., between images and sentences) to differences in computational efficiency; to carry out this reduction, he borrowed from computer science the concepts of data type and of data structure. I argue that, in the end, his account amounted to nothing more than characterizing representations by the fast operations on them. This analysis then allows me to assess what Simon’s approach actually achieves when transported from psychology to the study of scientific representations, as in Larkin and Simon (1987): it allows comparing, not representations in and of themselves, but rather the computational roles they play in particular problem-solving processes—that is, representations together with a particular way of using them. (shrink)

I encourage Johnson et al. to ground Conviction Narrative Theory in more detail in foundational, earlier decision-making research – first and foremost in Herbert Simon's work. Moreover, I wonder if and how further reflections about narratives could aid tackling two interrelated grand challenges of the decision sciences: To describe decision-making environments; to understand how people select among decision-strategies in environments.