Eva Knor is a third year PhD candidate based in Hamburg, Germany Contact
Under the joint supervision of Judith Möller and Katharina Kleinen-von Königslöw. My research sits at the intersection of political communication and human–computer interaction, with a focus on generative AI as a source of information. I study how users interact with these systems and how they respond to algorithmic failures and uncertainty, aiming to understand the broader implications for political communication and public engagement.
CHATGPT as news commander system
6social media platforms(Helberger et al., 2018), act as news retrieval systems and their effect on democratic citizenship. The debate mainly focuses on concerns over adverse effects of algorithmic content recommendations on exposure diversity. Critics argue that individually tailored news and information flows to user preferences risks narrowing the range of viewpoints and sources people encounter, either intentionally (filter bubble hypothesis) (Pariser, 2011), or because digital architectures of platforms make it easier for users to ignore or self-select attitude-congruent information environments (echo chamber hypothesis)(Cinelli et al., 2021). Others, however, highlight that empiricalevidence for significant reductions of exposure diversity is limited, suggesting that the impact of
ferfe
nweodjx k
TITLE: CHATGPT AS A NEWS RECOMMENDER SYSTEM6social media platforms(Helberger et al., 2018), act as news retrieval systems and their effect on democratic citizenship. The debate mainly focuses on concerns over adverse effects of algorithmic content recommendations on exposure diversity. Critics argue that individually tailored news and information flows to user preferences risks narrowing the range of viewpoints and sources people encounter, either intentionally (filter bubble hypothesis) (Pariser, 2011), or because digital architectures of platforms make it easier for users to ignore or self-select attitude-congruent information environments (echo chamber hypothesis)(Cinelli et al., 2021). Others, however, highlight that empiricalevidence for significant reductions of exposure diversity is limited, suggesting that the impact of algorithms may be less pronounced than often assumed(Haim et al., 2018; Möller et al., 2018). Research also increasingly highlights that more diversity does not necessarily translate into more democracy/ more democratic deliberation. In high-choice digital media systems, an abundance of heterogeneous content can overwhelm users, foster confusion, or even lead to political disengagement(Bawden & Robinson, 2009). Exposure to strongly opposing viewpoints can, under certain conditions, even intensify polarization rather than encourage deliberation(Bail et al., 2018). This suggests that the relationship between exposure diversity and democratic participation such as deliberation and discourse is not linear but conditional—dependent on users’ cognitive capacities, political orientations, and the architecture of recommendation systems that mediate access to information. Hence, the normative ideal of “more d
TITLE: CHATGPT AS A NEWS RECOMMENDER SYSTEM6social media platforms(Helberger et al., 2018), act as news retrieval systems and their effect on democratic citizenship. The debate mainly focuses on concerns over adverse effects of algorithmic content recommendations on exposure diversity. Critics argue that individually tailored news and information flows to user preferences risks narrowing the range of viewpoints and sources people encounter, either intentionally (filter bubble hypothesis) (Pariser, 2011), or because digital architectures of platforms make it easier for users to ignore or self-select attitude-congruent information environments (echo chamber hypothesis)(Cinelli et al., 2021). Others, however, highlight that empiricalevidence for significant reductions of exposure diversity is limited, suggesting that the impact of algorithms may be less pronounced than often assumed(Haim et al., 2018; Möller et al., 2018). Research also increasingly highlights that more diversity does not necessarily translate into more democracy/ more democratic deliberation. In high-choice digital media systems, an abundance of heterogeneous content can overwhelm users, foster confusion, or even lead to political disengagement(Bawden & Robinson, 2009). Exposure to strongly opposing viewpoints can, under certain conditions, even intensify polarization rather than encourage deliberation(Bail et al., 2018). This suggests that the relationship between exposure diversity and democratic participation such as deliberation and discourse is not linear but conditional—dependent on users’ cognitive capacities, political orientations, and the architecture of recommendation systems that mediate access to information. Hence, the normative ideal of “more d
