The project investigates the influence and effects of black-box algorithms on various stakeholders by qualitatively adn quantitaviely investigating the stakeholders' perceptions of algorithmic function. The project is part of the Princeton Dialogues on AI & Work , currently focusing on the analysis of online food delivery platforms.
The project analyzes how bias preexisting in text is diffused in models such as word embeddings and deep neural architectures. It also tries to develop automated methods for tracing and quantifying biases in text, such as sexism, racism and homophobia.
The project analyzes aspects of social media platforms and user behaviour for uncovering how political communication takes places. It studies phenomena as political microtargeting, automated and fake accounts, information diffusion, agenda-setting, polarization of public opinion, recommendation systems and political party behavior. Publications related to these topics can be found here: 1 2 3 4 5 6
Civic Machines is a blog that analyzes the interaction of individuals and technology under a cybernetic perspective. It illustrates the impact of sociotechnological ecosystems, such as social media platforms and automated decision making applications on individuals. Its aims to uncover issues, limits and possibilities of the ecosystems, while wants to promote their civic design and regulation. You can find the latest post here.
This project attempts to give an overview of the online political landscape in Germany. It is a live dashboard, which collects and analyzes political content from Twitter, Facebook and 40 online news media sources. The dashboard tries to: 1. Locate and present the trending political topics discussed in Germany online. 2. Provide insights on the attitudes and political interests of online partisan users. You can access it here.
During the German general election in 2017, there were coordinated attempts to disturb online public opinion. Disinformation campaigns used different social media platforms like Facebook and Twitter. They managed to infiltrate the online reporting of major German news outlets. Due to the distributed character of disinformation campaigns, these attempts were hardly noticed by the public, and their real dimension has not yet been revealed. The project investigates manipulation patterns and their effects on electoral results. It is part of the Social Science One partnership with Facebook and organized by the Social Science Research Council. More here.
Bias by us envisions a future of media diversity by understanding the bias of today. The project seeks to understand how the US media ecosystem reports on underrepresented minorities. By using natural language processing algorithms and data-intensive models, aims to uncover underlying stereotypes, associations, and modes of narration that media produces and reproduces when covering minority related events. The project is hosted at the center for civic media at the MIT media lab. More here.
The hackathon was co organized from the Bavarian Government and the Professorship of Political Data Science at the TU Munich for investigating ways to enhance citizen communication with the local government. Participants were asked to develop innovative AI-based models and applications towards that end. More here.
Munich Bot challenge was a competition organized by Stifterverband, the Munich Center for Internet Research and the Professorship of Political Data Science at the TU Munich with the aim to provide awareness on automated behaviour on social media. More here.
Fatum presents a variety of philosophical perspectives on current developments on Technology and Society and on multiple scientific disciplines. The magazine wants to stimulate discussions on multiple topics and initiate dialog between scientific fields. Fatum was developed by the MSc Program Philosophy of Science and Technology at the Technical University of Munich. More here.