
Homepage » Events calendar » The power of computer vision. A critical analysis
At 27 September 2023, Rosalie Waelen will defend her thesis 'The power of computer vision. A critical analysis'.
Description dissertation:
This dissertation introduces a new approach to AI ethics and applies this approach to discuss the ethical and societal implications of computer vision technology. Computer vision is the subfield of artificial intelligence (AI) that focuses on automizing the analysis of images and videos. Computer vision consists of different tasks, such as facial recognition and object tracking, and can serve a wide variety of purposes, including surveillance, business intelligence, healthcare, and convenience. As a result, computer vision is becoming omnipresent in our lives and societies. Since computer vision is a subfield of AI, the ethics of computer vision can be treated as a subfield of AI ethics. In order to analyze and evaluate the ethical and societal implications of computer vision, this dissertation offers a new approach to AI ethics: a critical approach. The critical approach to AI ethics is inspired by the Frankfurt School tradition of critical theory, in particular its focus on power relations and its emancipatory aim. The approach consists of a framework that allows one to identify normative issues in AI in terms of power and evaluate these issues in light of their impact on autonomy. The critical approach to AI ethics also encourages the use of critical theory’s concepts, theories, and methods, to discuss and transform AI’s ethical and societal implications. The critical approach to AI ethics combines ethical, social, and political concerns related to AI. Using the critical approach to AI ethics, the dissertation offers an overview of ethical and societal issues related to computer vision. Three issues are then analyzed in more detail – namely the constitutive effects of computer vision applications, the ways in which facial recognition applications can misrecognize people in a normative sense, and the exploitative nature of datafication practices.