AI and opinion formation: The danger of manipulation in democracy!

AI and opinion formation: The danger of manipulation in democracy!

On May 22, 2025, the Technical University of Berlin announced significant steps in research into the social dynamics of opinion formation. Under the leadership of Sebastian Pokutta, Vice President of the Zuse Institute Berlin and spokesman for the MATH+ Cluster of Excellence, the next funding phase will focus on the analysis of opinion formation through social media. The aim is to strengthen and protect democratic values, while highlighting artificial intelligence (AI) and optimization methods as key research areas. Pokutta and his team see a need to better understand how information is disseminated and the influence algorithms have on public opinion.

In recent years, the change and spread of opinions through social media has increased significantly. According to the Federal Agency for Civic Education, access to reliable information is seen as crucial for the functioning of democracy. The influence of social media has changed significantly, which has created social vulnerabilities. The German population shows declining trust in the media, falling from 48% in 2023 to 47% in 2024. Trust in the government has also decreased, from 47% to 42%.

The role of artificial intelligence

Artificial intelligence is playing an increasingly central role in political communication. Recent developments, particularly since the launch of ChatGPT at the end of 2022, have reignited the discussion about AI manipulation and disinformation. The ability of AI to generate automated content and specifically personalize it is identified as a major risk. Political actors specifically use algorithms to increase the visibility of certain narratives. This can be seen, for example, in the manipulative strategies that the AfD also uses to spread its messages.

In the planned MATH+ funding phase, Andrea Walther, another spokesperson for the project, attaches great importance to sustainability. Algorithms to increase efficiency should be developed in the areas of mobility and energy, e.g. by optimizing transport networks to reduce the power consumption of AI when learning.

Democratic challenges and recommendations for action

The spread of disinformation poses a significant challenge to democratic processes. Deep fakes and unreliable AI-generated content threaten democratic discourse and could undermine the conditions for an enlightened public. Social media based on complex algorithms maximize user engagement, leading to fragmentation of the information space and endangering social cohesion.

To counteract this development, concrete measures are required. Recommendations for action include the introduction of a code of conduct for political parties, public education campaigns and the promotion of a multi-stakeholder approach. Ensuring transparency via algorithms and access to user data are essential. Close cooperation between platform operators, politicians and civil society organizations is considered necessary in order to identify and counteract attempts at manipulation.

Overall, it shows that the challenges posed by AI and its far-reaching implications for democracy cannot be ignored. Developments in political communication require active participation and regulation to ensure that democratic values ​​are preserved even in a digitalized and algorithmically controlled world.