Refine
Has Fulltext
- yes (4)
Document Type
- Bachelor Thesis (2)
- Master's Thesis (2)
Is part of the Bibliography
- yes (4)
Many fields of science and business are currently undergoing technological changes. Possible future prospects and the impact of our ever increasing reliance on digital algorithms are key factors influencing this field of research. However, only few confront contemporary digitalization with socially unjustified effects. Therefore, this work focuses on the main research question: Can the transparency and fairness concerning the probabilistic distributions of computer based selections be ensured? Or, in contrary, do the current power ranks promote an amplification of bias found in training datasets that swiftly impact marginalized populations at scale. We are still a long way from intellectual and creative thinking computer networks. Our current notion of artificial intelligence (AI) is based on neural networks, which can emit individual "thought processes" based on historical data and homogeneous developer teams. This is not self contained thinking but merely a copy of human thought patterns, which may well be affected by discrimination, e.g. job selection,
credit-ratings, healthcare or other types of personal appraisals. Machine Learning (ML), so far, can not filter out individual human abilities based on historical data and discriminating coding.
This work discusses possible reasons for biased opinions of machine learning classification systems. Also, the social question focusing on gender roles and potential winners and losers through technological changes are explored. Finally, approaches for anti discriminatory use of AI are proposed.