Artificial intelligence (AI) is a key area of deliberation for companies, governments, academics, technologists, and civil society organisations.
Some stakeholders have focused on building normative ethical principles. Last week, the European Commission’s High Level Expert Group on Artificial Intelligence published ethical guidelines for Responsible AI. A few weeks prior, Google set up an AI ethics board to implement its ethical principles.
Other stakeholders have gravitated towards a technical discussion of how to build fair, accountable, and transparent AI systems. For instance, facial recognition technologies are being developed by companies and deployed by governments, and there are active attempts at making these systems fair and accurate. But it is becoming increasingly clear that these approaches are not enough. The European Commission’s High Level Expert Group on AI came under fire for being disproportionately swayed by industry interests, at the cost of civil society’s concerns, and for being an exercise in ethics washing. Google’s AI board was found to be constituted in opaque and problematic ways, with some of its members advocating ideas that were fundamentally incongruous with its own principles (the board has been dissolved since). The attempt to make facial recognition fair and accurate has been critiqued for missing fundamental problems – the technology prima facie challenges our ability to exercise rights, and more accurate systems only become more powerful systems of control and surveillance.
This report finds critical gaps in both ethical and technical approaches to AI systems. They are both lacking in enforcement mechanisms and definition: which institutions they invoke and how they empower individuals.
In this report, ARTICLE 19 considers the ethical and technical approaches in the field so far. We identify the contours and limitations of these parallel discussions, and then propose ways in which a human rights-based approach can constructively inform and enhance these efforts. The report makes recommendations for industry, states, and civil society.
For any media enquiries, please contact Vidushi Marda, Digital Programme Officer: firstname.lastname@example.org