Do you trust a robot journalist? The perceived credibility of AI generated articles in Germany

Previous research has shown that credibility in the media is a valuable asset and that audiences’ perception of a source's reliability can influence their choice of source. With the increasing implementation of automated journalism, which uses artificial intelligence (AI) and natural language generation (NLG), not only has news production changed fundamentally, but so has the assessment of the credibility of texts. This master's thesis examines how German readers evaluate the credibility of AI-generated news articles. This thesis used in-depth interviews with ten German-speaking participants of different ages and from diverse professional backgrounds. The assessed credibility was analysed through a qualitative phenomenological approach that incorporates experimental elements. The experimental component refers to the controlled exposure of participants to two distinct stimuli to elicit their lived experiences and interpretive responses. In detail, participants were confronted with two texts, one generated by AI software and one written by a journalist. The participants assessed credibility and shared their feelings, thoughts and expectations about what is needed to establish credibility in AI-generated articles. Through thematic analysis, participants’ reflections were systematically examined to identify recurring patterns, contrasts, and underlying reasoning in their credibility judgements. Thematic analysis shows that transparency in labelling, the genre of the text, emotional depth and readability, and pre-existing trust in the medium are crucial factors in perception. This work highlights the complexity of the credibility phenomenon and concludes that credibility is a dynamic, subjective phenomenon influenced by various factors. This study helps to strengthen the relationship between media and audiences in times of technological change.

Grantor dc:publisher

Stellenbosch : Stellenbosch University

Year dc:date.issued

2025

Author dc:creator

• Pinders, Lynn

Advisor dc:contributor.advisor

• Ncube, Meli

Repository record dc:identifier.uri

https://scholar.sun.ac.za/handle/10019.1/134779

OAI identifier oai:identifier

oai:scholar.sun.ac.za:10019.1/134779