Technological Biases and Human Prejudices: A Technoethical Challenge

Digital technologies are not objective towards people, as they and their effects are closely connected and interwoven with them. Thus, technology shapes humans, just as humans shape technology. The core objective of this paper is to demonstrate the compliance between technology and humans. Accordingly, the risk and the limitations that emerge on the technical and the human side of the analogous relationship can be made visible. In order to accomplish this goal, the component of bias and the prejudice of human understanding will be discussed first. Subsequently, it can be shown that both components are constitutive for human interactions with AI systems. In view of the insurmountability of technical risks which are attributable to biases and human limitations caused by prejudices in the process of understanding, it is therefore concluded that the risk groups - as they are, e.g., explained and discussed in the recent European AI Act - need to be expanded. The following risk groups have been distinguished: AI systems with unacceptable risk; AI systems with high risk; AI systems with limited risk; and those with the lowest risk. The analysis carried out reveals the implicit risk that every AI system entails in its interaction with humans and shows that an elementary risk group is required. It leads to the technoethical demand to aim for a human-machine attitude that allows a critical-constructive approach to AI systems. From a technoethical perspective, this human-machine attitude currently represents a research gap, which this paper attempts to address for the first time.