Medskapare eller redskap? : En studie av människa–AI-interaktion som informationsskapande fenomen och dess etiska implikationer

Sammanfattning

This thesis explores how interaction between humans and generative artificial intelligence can be understood as an entangled, co-constitutive practice of information production. Drawing on Karen Barad’s theory of agential realism and Luciano Floridi’s information ethics, the study examines not only how such interaction unfolds, but also what ethical implications it may carry. The empirical material consists of interviews with generative AI users and selected examples of their dialogues with generative AI systems. A diffractive analytical approach was applied, allowing theory and data to be read through one another in order to identify recurring patterns, tensions, and ethical questions. The analysis suggests that users tend to frame themselves as agents in control when discussing their own use of generative AI, while describing others’ use as more passive or uncritical. Moreover, the perceived risks of using generative AI, shape the users’ sense of trust, responsibility, and control. The thesis highlights how agency is distributed across human–AI interaction and how both epistemic and ontological boundaries are actively negotiated in practice. These findings offer a nuanced understanding of human–AI co-agency and its ethical implications, with relevance for future research, information literacy, and librarianship.