LLMM – Large Language
Media Manipulator

Demonstrator: Raising Awareness about Information Manipulation by Language Models

LLMM – Large Language Media Manipulator

The LLMM demonstrator has two main objectives: firstly, to explain language model technology understandably, and secondly, to raise awareness of how easily information can be manipulated. To this end, current news articles are selected and then automatically processed by language models to demonstrate various manipulation aspects. One example is a style change in an article, showing how differences in linguistic nuances and styles can influence the perception of a topic. Another example is a change of the political viewpoint in an article: a neutral article can be rewritten to reflect the viewpoint of a particular political group or party. This illustrates how easily information can be manipulated to support a specific agenda. With these demonstrations, we want to encourage a critical perception of consumed information and the verification of the sources. We also want to inform about the ability of language model technology to manipulate information and raise awareness of how easily human perception can be influenced.

How does the exhibit contribute to the topic of sustainability?

The demonstrator contributes to the sustainability of democracy by promoting an understanding of information manipulation, fostering critical thinking, and promoting informed decision-making in a democratic society, thereby strengthening the integrity and resilience of democratic processes.

Would you like to find out more about the research project?

Your contact person

Dr. Jonas Fegert

FZI projects on the topic of sustainability