The increasing use of AI in different societal contexts intensified the debate on risks, ethical problems and bias. Accordingly, promising research activities focus on debiasing to strengthen fairness, accountability and transparency in machine learning. There is, though, a tendency to fix societal and ethical issues with technical solutions that may cause additional, wicked problems. Alternative analytical approaches are thus needed to avoid this and to comprehend how societal and ethical issues occur in AI systems. Despite various forms of bias, ultimately, risks result from eventual rule conflicts between the AI system behavior due to feature complexity and user practices with limited options for scrutiny. Hence, although different forms of bias can occur, automation is their common ground. The paper highlights the role of automation and explains why deep automation bias (DAB) is a metarisk of AI. Based on former work it elaborates the main influencing factors and develops a heuristic model for assessing DAB-related risks in AI systems. This model aims at raising problem awareness and training on the sociotechnical risks resulting from AI-based automation and contributes to improving the general explicability of AI systems beyond technical issues.

Astonishing progress is being made in the field of artificial intelligence (AI) and particularly in machine learning (ML). Novel approaches of deep learning are promising to even boost the idea of AI equipped with capabilities of self-improvement. But what are the wider societal implications of this development and to what extent are classical AI concepts still relevant? This paper discusses these issues including an overview on basic concepts and notions of AI in relation to big data. Particular focus lies on the roles, societal consequences and risks of machine and deep learning. The paper argues that the growing relevance of AI in society bears serious risks of deep automation bias reinforced by insufficient machine learning quality, lacking algorithmic accountability and mutual risks of misinterpretation up to incrementally aggravating conflicts in decision-making between humans and machines. To reduce these risks and avoid the emergence of an intelligentia obscura requires overcoming ideological myths of AI and revitalising a culture of responsible, ethical technology development and usage. This includes the need for a broader discussion about the risks of increasing automation and useful governance approaches to stimulate AI development with respect to individual and societal well-being.

With social network sites (SNS) becoming a pervasive phenomenon, already existing conflicts with privacy are further intensified. As shown in this paper, online and (once) offline contexts increasingly conflate, thereby posing new challenges to the protection of the private sphere. SNS quickly evolve their features and challenge privacy preferences, often without user consent; ‘social graphs’ make social relations highly transparent; social plugins interconnect user traces from within and outside the SNS. As the large amounts of personal information available in SNS are processed with context-rich information, the individual’s informational self-determination is heavily strained. These data attract potential and real observers for behavioural advertising and also for profiling by security authorities. We argue that the emerging usage of SNS (social plugins, increasing role of biometrics and mobile computing) multiplies privacy challenges as all types of privacy become affected. This raises additional demand for public policy to foster privacy-by-design combined with awareness-raising mechanisms to improve informational self-determination.

Research and development as well as societal debates on the risks of artificial intelligence (AI) often focus on crucial but impractical ethical issues or on technocratic approaches to managing societal and ethical risks with technology. To overcome this, more practical, problem-oriented analytical perspectives on the risks of AI are needed. This article proposes an approach that focuses on a meta-risk inherent in AI systems: deep automation bias. It is assumed that the mismatch between system behavior and user practice in specific application contexts due to AI‑based automation is a key trigger for bias and other societal risks. The article presents the main factors of (deep) automation bias and outlines a framework providing indicators for the detection of deep automation bias ultimately triggered by such a mismatch. This approach intends to strengthen problem awareness and critical AI literacy and thereby create some practial use.

The present work aimed to assess the chondroprotective influence of chitosan and lecithin in a monoiodoacetate (MIA)-induced experimental osteoarthritis (OA) model. Forty male rats weighing 180–200 g were randomly distributed among the following five experimental groups (eight per group): control, MIA-induced OA, MIA-induced OA + chitosan, MIA-induced OA + lecithin, and MIA-induced OA + chitosan + lecithin. The levels of TNF-α, IL6, RF, ROS, and CRP, as well as mitochondrial markers such as mitochondrial swelling, cytochrome C oxidase (complex IV), MMP, and serum oxidative/antioxidant status (MDA level) (MPO and XO activities) were elevated in MIA-induced OA. Also, SDH (complex II) activity in addition to the levels of ATP, glutathione (GSH), and thiol was markedly diminished in the MIA-induced OA group compared to in control rats. These findings show that mitochondrial function is associated with OA pathophysiology and suggest that chitosan and lecithin could be promising potential ameliorative agents in OA animal models. Lecithin was more effective than chitosan in ameliorating all of the abovementioned parameters.

National electronic identity (e-ID) card schemes and electronic identity management systems (e-IDMS) in Europe are characterised by considerable diversity. This contribution analyses the creation of a national e-IDMS in Austria with the aim of improving our understanding of the reasons behind the genesis of particular designs of national e-IDMS. It seeks to explain how the system’s specific design evolved and which factors shaped its appearance. Being part of a comparative four country study, a common theoretical framework is employed to allow for a comparison of national e-IDMS in Austria, Belgium, Germany and Spain. It combines the approach of actor-centred institutionalism and the concept of path dependence in order to analyse the innovation process and to explain resulting key characteristics of the e-IDMS in Austria: a technology-neutral system with multiple tokens; an ID model based on the Central Register of Residents; a privacy concept using sector-specific personal identifiers. It is shown that innovation process and outcome are not only shaped by specific actor constellations dominated by strategic e-government bodies, but also by path dependence at three levels: technological, institutional and organisational.