Content

StereOFF ‒ automated identification and analysis of gender stereotypes - Civic Innovation Platform

StereOFF ‒ automated identification and analysis of gender stereotypes

Gender stereotypes are one of the main causes of gender inequalities. The German Research Center for Artificial Intelligence (DFKI) and Didactic Innovations GmbH want to tackle this challenge with their idea, which is about the automated identification of gender stereotypes and associated implicit discrimination in learning materials for education and further training. This will contribute to inclusive education and training and thus counteract challenges resulting from gender inequalities, such as the gender pay gap and the shortage of skilled workers. Thanks to their technological basis, we will be able to transfer these analyses to a wide range of domains in the long term and thus combat inequalities at different levels of society.

What makes you a strong team? 

We are a strong team because our competencies complement each other: didactic knowledge meets AI expertise for innovative educational solutions. Shared objectives, rapid implementation of research into practice and expanded partnerships enable efficient, sustainable solutions. This cooperation strengthens our opportunities and enables us to successfully implement challenging projects and make a positive contribution to society.

Explain your idea in three sentences. 

Our tool StereOFF identifies and automates gender stereotypes in education and training materials and makes users aware of implicit discrimination. The analyses are based on publicly available text corpora (e.g. film transcripts) that have a natural gender bias and have shaped our gender socialisation. The aim of StereOFF is to dismantle gender stereotypes, especially those that restrict women’s and men’s choice of careers.

What makes your idea special? 

By using data from popular cultural media with a natural gender bias (e.g. films, social media), our analytics have a high transfer potential. While previous approaches have often been limited to the analysis of job adverts and thus to a few words, StereOFF can be adapted to an inexhaustible number of application domains in the long term by using publicly available text corpora and already trained language models.

What’s next? 

Firstly, initial analyses will be carried out with the film data in order to demonstrate the feasibility of the project. To this end, a reduced data set is extracted and used to evaluate various AI models and adapt them to the specific use case. This is followed by an initial evaluation to show the benefits of the project idea. The aim of this preparatory work is to transform the project idea into a fully fledged project in order to enable overall implementation.

Contact