Innovation vs. Safety: A critical examination of regulatory approaches to artificial intelligence

The functioning of society has been forever changed by the creation of artificial intelligence (AI); what was once a futuristic and unrealistic invention of science fiction now pervades our everyday lives in inconceivable ways in certain parts of the world. In the UK, we are almost always in direct contact with an AI-based system, whether this be via the smart phones we keep in our pockets, when competing our weekly shopping in-store or online, or when we are deciding what to watch at the end of a busy day. The ubiquitous state of this technology gives rise to both curiosity and concern. Whilst we may acknowledge that this technology streamlines and simplifies several of our daily activities and tasks, it is without question that there is an underlying unease associated with AI use due to its capricious nature. It is clear that efforts to develop increasingly sophisticated AI and implement these systems at any given opportunity will not cease, therefore we must give justifiable consideration to the regulation of this technology, which at present is considerably sparse. This thesis therefore proposes an innovative approach to AI regulation, and in doing so examines in detail the various risks associated with AI use (using bias and discrimination within AI systems as a case study) and scrutinises the regulatory proposals for governing this technology presented by a variety of national, regional, and international bodies and states, in order to assess the current state of global readiness for AI governance. Overall, this thesis purports that there is a significant lack of research in reasonable and workable governance measures for AI, and as such considerable work must be undertaken in this space.To this effect, a predominantly doctrinal and comparative approach is taken to this interdisciplinary project, and the research undertaken within this thesis contributes to the literature in several ways. Firstly, the thesis presents an internationally comparative analysis of AI regulatory proposals. This analysis examines, in depth, the various weaknesses of these regimes and proposes reasonable amendments to such. These proposals are made with a view to being robust, realistic and workable, and therefore have the capacity to be truly impactful. Secondly, this thesis features a detailed evaluation of the issues and key features necessary within any regulatory regime specifically targeted at governing modern technologies. An examination of this kind is lacking in current scholarship in this area, and so the work undertaken in this thesis contributes to this gap in the literature. Finally, by presenting reasonable recommendations, a workable proposed framework for harmonisation, including a rights-based impact assessment unique to this work, this thesis makes another original contribution to research in this space. This contribution is informed by the findings in the initial chapters of this thesis and encourages policymakers and professionals in this space to think innovatively about how we regulate AI