AutoVision: Complementing designer judgment and exploring cognitive limitations in responsible autonomous vehicle design

Abstract Responsible design requires anticipating uncertain, systemic and distant consequences, despite cognitive limitations in reasoning beyond familiar practices and immediate effects. This paper presents AutoVision, an anticipatory design platform supporting the generation and evaluation of autonomous vehicle (AV) concepts. It combines the Study–Method–Impact (SMI) model, which structures scientific evidence on long-term socioeconomic and environmental impacts, with the Representation–Usage–Impact (RUI) model, which captures expert knowledge on social and ethical consequences. Through its Advisor and Impact Guesser modes, AutoVision links impact-oriented objectives to design directions and concepts to traceable impact pathways. We evaluated the platform with eight senior designers from an automotive group to examine its relevance and the cognitive limitations affecting responsible concept evaluation. Participants created and evaluated AV concepts, including one generated through AutoVision without knowing its origin. The AutoVision-generated concept was perceived as at least as convincing as the designer-generated alternatives. Participant judgments diverged from the platform’s evidence-supported evaluations, particularly on environmental and ethical issues, while their justifications covered a narrower and less nuanced range of consequences. These exploratory findings show how anticipatory support can complement designer judgment, foster reflexivity, and help address cognitive limitations in responsible concept design without claiming to predict future impacts.