Artificial Intelligence (AI)-based systems are widely employed nowadays to make decisions that have far-reaching impact on individuals and society. Their decisions might affect everyone, everywhere, and anytime, entailing concerns about potential human rights issues. Therefore, it is necessary to move beyond traditional AI algorithms optimized for predictive performance and embed ethical and legal principles in their design, training, and deployment to ensure social good while still benefiting from the huge potential of the AI technology. The goal of this survey is to provide a broad multidisciplinary overview of the area of bias in AI systems, focusing on technical challenges and solutions as well as to suggest new research directions towards approaches well-grounded in a legal frame. In this survey, we focus on data-driven AI, as a large part of AI is powered nowadays by (big) data and powerful machine learning algorithms. If otherwise not specified, we use the general term bias to describe problems related to the gathering or processing of data that might result in prejudiced decisions on the bases of demographic features such as race, sex, and so forth.This article is categorized under:

The ubiquity of digital devices and the increasing intensity of users’ interactions with them create vast amounts of digital trace data. Companies use these data to optimize their services or products, but these data are also of interest to researchers studying human behavior. As most of these data are owned by private companies and their collection requires adherence to their terms of service, research with digital trace data often entails some form of public-private partnership. Private companies and academic researchers each have their own interests, some of which are shared, while others may conflict. In this article, we explore different types of private-public partnerships for research with digital trace data. Based on general considerations and particular experiences from a research project with linked digital trace data, we propose strategies for identifying and productively negotiating both shared and conflicting interests in these relationships.

Sharing social media research datasets allows for reproducibility and peer-review, but it is very often difficult or even impossible to achieve due to legal restrictions and can also be ethically questionable. What is more, research data repositories and other research infrastructure and research support institutions are only starting to target social media researchers. In this paper, we present a practical solution to sharing social media data with the help of a social science data archive. Our aim is to contribute to the effort of enhancing comparability and reproducibility in social media research by taking some first steps towards setting standards for sustainable data archiving. We present a showcase for sharing social media data with the example of a big dataset containing geotagged tweets (several months of continued geotagged tweets from the United States from 2014 and 2015; nearly half a billion tweets in total) through a research data archive. We provide a general background to the process of long-term archiving of research data. After some consideration of the current obstacles for sharing and archiving social media data, we present our solution of archiving the specific dataset of geotagged tweets at the GESIS Data Archive for the Social Sciences, a publicly funded German data archive for secure and long-term archiving of social science data. We archived and documented tweet IDs and additional information to improve reproducibility of the initial research while also attending to ethical and legal considerations, and taking into account Twitter's terms of service in particular.

"More and more researchers want to share research data collected from social media to allow for reproducibility and comparability of results. With this paper we want to encourage them to pursue this aim - despite initial obstacles that they may face. Sharing can occur in various, more or less formal ways. We provide background information that allows researchers to make a decision about whether, how and where to share depending on their specific situation (data, platform, targeted user group, research topic etc.). Ethical, legal and methodological considerations are important for making this decision. Based on these three dimensions we develop a framework for social media sharing that can act as a first set of guidelines to help social media researchers make practical decisions for their own projects. In the long run, different stakeholders should join forces to enable better practices for data sharing for social media researchers. This paper is intended as our call to action for the broader research community to advance current practices of data sharing in the future." (author's abstract

No abstract