Professor Lee Bygrave
We live in a world where automated decisional systems based on machine learning (ML) and ‘Big Data’ increasingly govern our behaviour. These systems promise a range of benefits, yet they also throw up a congeries of challenges, not least for our ability as humans to understand their logic and ramifications. This presentation considers such challenges through the prism of ‘cognitive sovereignty’ – a notion coined by Ulrich Beck in his famous work Risikogesellschaft (1986).
For the purposes of the presentation, cognitive sovereignty essentially denotes our moral and legal interest in being able to comprehend our environs and ourselves. The presentation argues that focus on cognitive sovereignty fills a blind spot in scholarship and policy discourse on the challenges arising from increasing use of ML-enhanced decisional systems. Not only is the notion an important constituent for an overarching conceptual framing of these challenges, it is also vital for grounding normative claims for greater explicability of machine processes. Further, the presentation assesses briefly the role of law, focusing on the provisions of data protection law that specifically concern automated decision-making. It shows that data protection law provides considerable but limited support for our cognitive sovereignty. At the same time, the application of legal norms operating with broad-brush criteria of proportionality, balance, and fairness carries a danger of ‘grey-box’ decision-making whereby the ‘black boxes’ of ML-enhanced decisional systems are assessed according to woolly, relatively subjective notions of propriety that are also rather opaque.
No comments:
Post a Comment