The research project “Generative Artificial Intelligence for Information Navigation“, funded by the German Federal Ministry of Education and Research (BMBF), investigated to what extent, for what purposes and for what reasons the German population uses generative artificial intelligence in applications such as ChatGPT or Google Gemini. The findings can be downloaded as a working paper.
Reiss, Michael V.; Knor, Eva Luise; Stöwing, Ezra; Merten, Lisa; Möller, Judith (2025): Zwischen Neugier und Skepsis: Nutzung und Wahrnehmung generativer KI zur Informationssuche in Deutschland [Between Curiosity and Skepticism: Use and Perception of Generative AI for Information Search in Germany]. Hamburg: Verlag Hans-Bredow-Institut, March 2025 (Working Papers of the Hans-Bredow-Institut | Project Results No. 76), https://doi.org/10.21241/ssoar.100907
Study Overview and Summary
The rapid spread of generative artificial intelligence (AI) through applications such as ChatGPT or Google Gemini is fundamentally changing the media and information landscape. The technology is often discussed critically because of its susceptibility to errors in the context of information gathering. It is particularly problematic that even wrong answers often seem plausible and are therefore difficult to recognize as incorrect by people without in-depth expertise in the respective subject area.
However, since it is still largely unclear to what extent, for what purposes, and for what reasons the German population uses generative AI, it is also difficult to assess the relevance of the topic. This is where the research project “Generative Artificial Intelligence for Information Navigation” comes in, which aims to better understand the dissemination, use and acceptance of generative AI in Germany. In addition to general motives for use and attitudes, the project focused in particular on the question of whether and how generative AI is used for political information purposes or as a substitute for news in the context of the 2024 European elections, as false information is particularly problematic in this context.
A total of 1,461 people in Germany participated in the representative survey in May and June 2024.
Key findings
- Generative AI is already used by nearly 44 percent of the online population. Usage is strongly correlated with age. In the 16-19 age group, nearly 96 percent already use generative AI. In the 60-69 age group, the figure is just over 18 percent.
- The majority of users rarely use generative AI; the technology is not yet integrated into everyday life. However, this is not the case for young people in an educational context. Here, generative AI is already used frequently by a majority, which makes it necessary to pay more attention to the risks and opportunities of its use.
- The main reason for not using generative AI is a lack of personal need. In addition, comprehension problems, such as not knowing where or how generative AI can be used, are particularly relevant for older people. If generative AI is to be made accessible to everyone, these barriers should be specifically reduced.
- The use of generative AI for political information or as a substitute for news is currently not very widespread. People who use generative AI for such purposes are also more likely to use traditional news media, which mitigates concerns that generative AI could replace journalism.
Given the dynamic development and increasing integration of generative AI into different areas of life, it is still necessary to monitor the use of this technology in order to realistically assess its opportunities and risks in the future.
The most important findings can also be found in two videos and a podcast.
- Part 1 of the video on YouTube is available here: https://youtu.be/5vLaZMqznV0
- Part 2 of the video on YouTube is available here: https://youtu.be/o6__gz4TTfY
- And for those who prefer to listen to a podcast on the topic: In BredowCast #97 “Verbreitung, Nutzung und Akzeptanz generativer KI-Systeme in der deutschen Bevölkerung [Distribution, Use and Acceptance of Generative AI Systems in the German Population]”, Eva Knor and Michael Reiss discuss their research findings and explain why it makes a difference whether I ask a generative AI how long to boil an egg or which party stands for what, why ChatGPT “knows” that a traffic light is red, yellow and green and not blue, and why a Google search is sometimes a better advisor.