idw – Informationsdienst Wissenschaft

Nachrichten, Termine, Experten

Grafik: idw-Logo
Science Video Project
idw-Abo

idw-News App:

AppStore

Google Play Store



Instance:
Share on: 
01/10/2024 14:15

The Achilles Heel of Artificial Intelligence: Why Discrimination Remains an Unresolved Problem

Carolin Höll Hochschulkommunikation
Duale Hochschule Baden-Württemberg

    Report: Baden-Wuerttemberg Cooperative State University

    A recent study by the DHBW Stuttgart at the Service Management Study Center (ZMM) investigated the ability of Artificial Intelligence (AI) to recognize discriminatory content in images and advertisements, showing both impressive progress as well as existing limitations.

    In this comprehensive study, AI was confronted with a variety of images and advertisements and asked to evaluate them. This included 60 advertisements, among others, that the German Advertising Council had recently criticized. The results show that AI has an astonishing ability to identify discriminations in advertisements with impressive accuracy (F1 Score: 0.949). As a result, the AI generally assessed the advertisements criticized by the German Advertising Council as potentially discriminatory. At the same time, in most cases, non-discriminatory advertisements did not receive such a warning.

    This is particularly impressive considering that just ten years ago, AI had significant
    difficulties in correctly classifying objects depicted in an image. In the meantime, and thanks to millions of images, AI has learned to largely accurately distinguish between a dog and a cat. The rapid progress of AI raises the question of where its current limits lie. "We wanted to know to what extent AI recognizes discriminatory behaviour when it is only presented with an advertisement and asked to evaluate it," explains student Helen Beckers about the approach of the study.

    Algorithmic Discrimination
    This is of particular relevance in light of the massive increase in discrimination by algorithms that can disadvantage people based on gender, religion, ideology, racism, or origin. This aspect is becoming more important as discrimination by algorithms is increasingly becoming a serious problem affecting various areas such as application processes, credit allocation, medicine, and the calculation of the recidivism probability of offenders.

    ChatGPT's Ability to Uncover Sexualization and Stereotypes
    The insight that ChatGPT can detect sexualization and stereotypical thinking is particularly revealing. A modified advertisement with swapped gender roles showed that AI can also identify discrimination in reversed situations. "This different evaluation by ChatGPT in the two scenarios highlights the ability to identify discrimination even in reversed situations," says student Sven Peter, sharing another finding from the study.

    Limits and Challenges in Discrimination Recognition by AI
    However, AI reached its limits in identifying other forms of discrimination, such as objectification, disrespect, and abuse of power. "The results of the study underscore the need to further develop AI systems to recognize discrimination more effectively and prevent it," reports student Marius Funk. The study raises important questions about how AI technologies can be used in the future to combat discrimination in various areas and promote equality.

    Service Management Study Center
    The study was conducted under the leadership of Prof. Dr. U. Bucher at the Service Management Study Center (ZMM) at the DHBW Stuttgart. The Service Management Study Center prepares students for responsible business roles in innovative and customer-oriented projects as well as in various functional and Service areas of service companies.

    DHBW Stuttgart
    The Baden-Wuerttemberg Cooperative State University Stuttgart, with around 8,000 students in bachelor's and master's degree programs, is one of the largest higher education institutions in the Stuttgart and Upper Neckar regions. In cooperation with around 2,000 selected companies and social institutions, it offers more than 60 nationally and internationally recognized, work-integrated study programs in the fields of business, engineering, social work, and health. Bachelor students alternate every three months between the university and the training company, their dual partner. In this way, they gain professional experience during their studies. Financial independence, secure job market opportunities due to exceptionally high recruitment rates, small class sizes, and an international orientation offer tremendous advantages to students. Beyond the bachelor's degree, DHBW offers a dual master's program.
    More information at: www.dhbw-stuttgart.de


    Contact for scientific information:

    Prof. Dr. Ulrich Bucher
    Baden-Wuerttemberg Cooperative State University Stuttgart (DHBW Stuttgart)
    Theodor-Heuss-Str. 2
    70174 Stuttgart
    Germany
    E-Mail: ulrich.bucher@dhbw-stuttgart.de
    mobile: +49 177 / 2307 364


    Images

    Criteria of this press release:
    Business and commerce, Journalists, Scientists and scholars
    Economics / business administration, Information technology, Media and communication sciences, Social studies
    transregional, national
    Research results, Transfer of Science or Research
    English


     

    Help

    Search / advanced search of the idw archives
    Combination of search terms

    You can combine search terms with and, or and/or not, e.g. Philo not logy.

    Brackets

    You can use brackets to separate combinations from each other, e.g. (Philo not logy) or (Psycho and logy).

    Phrases

    Coherent groups of words will be located as complete phrases if you put them into quotation marks, e.g. “Federal Republic of Germany”.

    Selection criteria

    You can also use the advanced search without entering search terms. It will then follow the criteria you have selected (e.g. country or subject area).

    If you have not selected any criteria in a given category, the entire category will be searched (e.g. all subject areas or all countries).