idw – Informationsdienst Wissenschaft
Nachrichten, Termine, Experten
Nachrichten, Termine, Experten
Joint study of the Mainz University of Applied Sciences and Johannes Gutenberg University Mainz shows that AI-generated content is rated as having higher clarity and appeal
In a time when the Internet has become the main source of information for many people, the credibility of online content and its sources has reached a critical tipping point. This concern is intensified by the proliferation of generative artificial intelligence (AI) applications such as ChatGPT and Google Bard. Unlike traditional platforms such as Wikipedia, which are based on human-generated and curated content, these AI-driven systems generate content autonomously - often with errors. A recently published study, jointly conducted by researchers from the Mainz University of Applied Sciences and Johannes Gutenberg University Mainz (JGU), is dedicated to the question of how users perceive the credibility of human-generated and AI-generated content in different user interfaces. More than 600 English-speaking participants took part in the study.
As Professor Martin Huschens, Professor for Information Systems at the Mainz University of Applied Sciences and one of the authors of the study, emphasized: "Our study revealed some really surprising findings. It showed that participants in our study rated AI-generated and human-generated content as similarly credible, regardless of the user interface." And he added: "What is even more fascinating is that participants rated AI-generated content as having higher clarity and appeal, although there were no significant differences in terms of perceived message authority and trustworthiness – even though AI-generated content still has a high risk of error, misunderstanding, and hallucinatory behavior."
The study sheds light on the current state of perception and use of AI-generated content and the associated risks. In the digital age, where information is readily available, users need to apply discernment and critical thinking. The balance between the convenience of AI-driven applications and responsible information use is crucial. As AI-generated content becomes more widespread, users must remain aware of the limitations and inherent biases in these systems.
Professor Franz Rothlauf, Professor of Information Systems at Johannes Gutenberg University Mainz, added: "The study results show that – in the age of ChatGPT – we are no longer able to distinguish between human and machine language and text production. However, since AI does not 'know', but relies on statistical guessing, we will need mandatory labeling of machine-generated knowledge in the future. Otherwise, truth and fiction will blur and people cannot tell the difference." It remains a task of science communication and, not least, a social and political challenge to sensitize users to the responsible use of AI-generated content.
Image:
https://download.uni-mainz.de/presse/03_wirtschaft_chatgpt_eingabemaske.jpg
ChatGPT input screen
photo/©: Dominik Sobania and Louisa Schwarz
Related links:
• https://en.wi.bwl.uni-mainz.de/ – Information Systems and Business Administration at the JGU Gutenberg School of Management and Economics
• https://en.wiwi.uni-mainz.de/ – Gutenberg School of Management and Economics (GSME) at JGU
Dr. Dominik Sobania
Information Systems and Business Administration
Gutenberg School of Management and Economics
Johannes Gutenberg University Mainz
55099 Mainz, GERMANY
phone: +49 6131 39-29452
e-mail: dsobania@uni-mainz.de
https://en.wi.bwl.uni-mainz.de/dominik-sobania/
M. Huschens et al., Do You Trust ChatGPT? – Perceived Credibility of Human and AI-Generated Content, arXiv, 5 September 2023
DOI: 10.48550/arXiv.2309.02524
https://arxiv.org/abs/2309.02524
ChatGPT input screen
photo/©: Dominik Sobania and Louisa Schwarz
Merkmale dieser Pressemitteilung:
Journalisten, Lehrer/Schüler, Studierende, jedermann
Informationstechnik, Pädagogik / Bildung, Politik, Recht, Wirtschaft
überregional
Forschungsergebnisse, Wissenschaftliche Publikationen
Englisch
ChatGPT input screen
photo/©: Dominik Sobania and Louisa Schwarz
Sie können Suchbegriffe mit und, oder und / oder nicht verknüpfen, z. B. Philo nicht logie.
Verknüpfungen können Sie mit Klammern voneinander trennen, z. B. (Philo nicht logie) oder (Psycho und logie).
Zusammenhängende Worte werden als Wortgruppe gesucht, wenn Sie sie in Anführungsstriche setzen, z. B. „Bundesrepublik Deutschland“.
Die Erweiterte Suche können Sie auch nutzen, ohne Suchbegriffe einzugeben. Sie orientiert sich dann an den Kriterien, die Sie ausgewählt haben (z. B. nach dem Land oder dem Sachgebiet).
Haben Sie in einer Kategorie kein Kriterium ausgewählt, wird die gesamte Kategorie durchsucht (z.B. alle Sachgebiete oder alle Länder).
