Western societies are marked by diverse and extensive biases and inequality that are unavoidably embedded in the data used to train machine learning. Algorithms trained on biased data will, without intervention, produce biased outcomes and increase the inequality experienced by historically disadvantaged groups. Recognising this problem, much work has emerged in recent years to test for bias in machine learning and AI systems using various fairness and bias metrics. Often these metrics address technical bias but ignore the underlying causes of inequality. In this paper we make three contributions. First, we assess the compatibility of fairness metrics used in machine learning against the aims and purpose of EU non-discrimination law. We show that the fundamental aim of the law is not only to prevent ongoing discrimination, but also to change society, policies, and practices to ‘level the playing field’ and achieve substantive rather than merely formal equality. Based on this, we then propose a novel classification scheme for fairness metrics in machine learning based on how they handle pre-existing bias and thus align with the aims of non-discrimination law. Specifically, we distinguish between ‘bias preserving’ and ‘bias transforming’ fairness metrics. Our classification system is intended to bridge the gap between non-discrimination law and decisions around how to measure fairness in machine learning and AI in practice. Finally, we show that the legal need for justification in cases of indirect discrimination can impose additional obligations on developers, deployers, and users that choose to use bias preserving fairness metrics when making decisions about individuals because they can give rise to prima facie discrimination. To achieve substantive equality in practice, and thus meet the aims of the law, we instead recommend using bias transforming metrics. To conclude, we provide concrete recommendations including a user-friendly checklist for choosing the most appropriate fairness metric for uses of machine learning and AI under EU non-discrimination law.
The shift towards the use of algorithms in business has transformed merchant–consumer interactions. Products and services are increasingly tailored for consumers through algorithms that collect and analyze vast amounts of data from interconnected devices, digital platforms, and social networks. While traditionally merchants and marketeers have utilized market segmentation, customer demographic profiles, and statistical approaches, the exponential increase in consumer data and computing power enables them to develop and implement algorithmic techniques that change consumer markets and society as a whole. Algorithms enable targeting of consumers more effectively, in real-time, and with high predictive accuracy in pricing and profiling strategies. In so doing, algorithms raise new theoretical considerations on information asymmetry and power imbalances in merchant–consumer interactions and multiply existing biases and discrimination or create new ones in society. Against this backdrop of the concentration of algorithmic decision-making in merchants, the traditional understanding of consumer protection is overdue for change, and normative debate about fairness, accountability, and transparency and interpretive considerations for non-discrimination is necessary. The theory that notice and choice in data protection laws and consumer protection laws are sufficient in an algorithmic era is inadequate, and countervailing consumer empowerment is necessary to balance the power between merchants and consumers. While legislative activity and regulation have conceivably increased consumer-empowerment, such measures may provide a limited or unclear response in the face of the transformative nature of algorithms. Instead, policy makers should consider responsible algorithmic code and other proposals as potentially effective responses in the analysis of socio-economic dimensions of algorithms in business.
This article explores the use of artificial intelligence to help define the current test for copyright infringement. Currently, the test for copyright infringement requires the jury or a judge to determine whether the parties’ works are “substantially similar” to each other from the vantage point of the “ordinary observer.” This “substantial similarity” test has been criticized at almost every level due to its inconsistent nature. Artificial intelligence has evolved to the point where it can be used as a tool to resolve many of the current issues associated with the “substantial similarity” test. Specifically, courts would no longer have to rely on a battle of the experts or the use of lay observers to determine if a work is substantially similar to another work. Using a new test based on the “ordinary AI observer” copyright infringement can be established using a means that is both less biased and more fact driven while giving alleged infringers a means by which to check ex ante if their work could be infringing.
National cybersecurity plays a crucial role in protecting our critical infrastructure, such as telecommunication networks, the electricity grid, and even financial transactions. Most discussions about promoting national cybersecurity focus on governance structures, international relations, and political science. In contrast, this Article proposes a different agenda and one that promotes the use of innovation mechanisms for technological advancement. By promoting inducements for technological developments, such innovation mechanisms encourage the advancement of national cybersecurity solutions. In exploring possible solutions, this Article asks whether the government or markets can provide national cybersecurity innovation. This inquiry is a fragment of a much larger literature on various innovation policy options (including patents, prizes, grants, and research and development tax credits). It requires determining whether national cybersecurity is a public good and an examination of market failure and government failure. Along the way, it draws on a property-liability rules theoretical framework to argue that the patent system’s invention secrecy restrictions and government patent use are ineffective for national cybersecurity innovation. On a normative level, the interface between government intervention and markets presents innovation mechanisms for national cybersecurity. Turning to prescriptions, expansion of prizes should rapidly promote national cybersecurity innovation, and reciprocal public–private research and development interactions should gradually multiply knowledge spillovers.
