How Artificial Intelligence recommendation systems impact human decision-making

This research employed semiotic approaches to address the communication mechanisms underlying AI-mediated environments for their users, exploring how Artificial Intelligence recommender systems (AIRS) as a tool can aid and impact users’ decisions, actions and behaviours. This research aimed to fill the existing gap in understanding the influence of AI on its users, focusing specifically on the impact of AIRS on individuals engaging with digital environments such as social media. It addressed the effects of daily interactions with AI on individuals across cognitive, physiological, emotional, axiological and pragmatic dimensions, with the subsequent goal of understanding how these effects shape social practices and culture. Theoretical and empirical findings are closely connected and yield for semiotic and qualitative approaches in studying AI-human interactions and HCI in general, as well as give highlights on the current situation within the field. The main contribution of this work include theoretical and empirical findings. Theoretical findings allow introduction of a semiotic methodology for modelling and designing user-centred AI systems to enhance understanding of HCI and their impact on individual, social and cultural dimensions. Empirical findings suggest that AIRS can impact users’ learning process through AI-mediated 1) categories to interpret daily experiences. 2) Users of different age categories can find different affordances reinforcing AI tools to 3) receive stimuli that impact desired states at all levels (from body reactions to individual and social behaviours). Moreover, the development and regulations of AI lead to managing the trustworthiness of AI, meaning that users rely on AI through the bond of trust 4) accepting values and practices and 5) adapting to AI-mediated environments and digital representations of immediate physical environment, making it part of daily behaviors and practices. This research addresses how AIRS exercise their capacity to redefine behavioural patterns and societal and cultural norms. Semiotic perspectives in this research helped to align existing knowledge in psychology, decision-making modelling, behavioural studies, cultural studies, to highlight how HCI can be understood and researched today from a user’s perspective. It also helped overcome the gap in technology-oriented decision-making when research is mainly focused on how to create more advanced solutions and improve current state of art in AI. Instead, it proposes a user oriented decision-making focused on human needs and capacities to adopt these technologies in most optimal ways. Overall, this research underscores the need for more nuanced AI systems that respect and enhance user agency and cultural diversity, thereby fostering environments where technology serves as a facilitator of enriching human experiences rather than dictating them. The findings of this research call towards more holistic and human-centred approaches in AI development and applications.