Julia Graefe
The use of artificial intelligence (AI) in everyday products increases and thereby also the interest on human-AI interaction. Research on user-adaptive systems still lacks uniform terminology and common definitions. To fill this gap, a taxonomy of user-adaptive systems, the Levels of Adaptive Sensitive Responses (LASR), has been proposed in previous research. The LASR levels are a first step towards a uniform classification of user-adaptive systems, but still need to be validated in terms of comprehensiveness and accuracy. This paper presents an evaluation of the LASR taxonomy through its application on technical products, to examine whether the taxonomy is complete and applicable in a meaningful way. We used a four steps analysis approach, covering the definition of classification criteria, two research iterations on suitable user-adaptive product features, and the analysis to which degree these features fulfill the classification criteria. As a result, our analysis shows that nine out of eleven analyzed product features can clearly be categorized into one of the five levels of LASR. Two features showed characteristics of two different levels which reveals that the levels are not mutually exclusive but can be combined within one product feature.