Ngozi Okidegbe (Yeshiva University – Benjamin N. Cardozo School of Law) has posted “When They Hear Us: Race, Algorithms and The Practice of Criminal Law” (Kansas Journal of Law & Public Policy, Vol. 29, 2020) on SSRN. Here is the abstract:

We are in the midst of a fraught debate in criminal justice reform circles about the merits of using algorithms. Proponents claim that these algorithms offer an objective path towards substantially lowering high rates of incarceration and racial and socioeconomic disparities without endangering community safety. On the other hand, racial justice scholars argue that these algorithms threaten to entrench racial inequity within the system because they utilize risk factors that correlate with historic racial inequities, and in so doing, reproduce the same racial status quo, but under the guise of scientific objectivity.

This symposium keynote address discusses the challenge that the continued proliferation of algorithms poses to the pursuit of racial justice in the criminal justice system. I start from the viewpoint that racial justice scholars are correct about currently employed algorithms. However, I advocate that as long as we have algorithms, we should consider whether they could be redesigned and repurposed to counteract racial inequity in the criminal law process. One way that algorithms might counteract inequity is if they were designed by most impacted racially marginalized communities. Then, these algorithms might counterintuitively benefit these communities by endowing them with a democratic mechanism to contest the harms that the criminal justice system’s operation enacts on them.

Aislinn Kelly-Lyth (Harvard Law School, University of Cambridge – Faculty of Law) has posted “Challenging Biased Hiring Algorithms” (Oxford Journal of Legal Studies (March 2021) on SSRN. Here is the abstract:

Employers are increasingly using automated hiring systems to assess job applicants, with potentially discriminatory effects. This paper considers the effectiveness of EU-derived laws, which regulate the use of these algorithms in the UK. The paper finds that while EU data protection and equality laws already seek to balance the harms of biased hiring algorithms with the benefits of their use, enforcement of these laws in the UK is severely limited in practice. One significant problem is transparency, and this problem is likely to exist across the EU. The paper therefore recommends that data protection impact assessments, which must be carried out by all employers using automated hiring systems in the EU or UK, should be published in redacted form. Mandating, and in the short term incentivising, such publication would enable better enforcement of rights which already exist.

Patric Reinbold (University of Wisconsin – Madison, School of Law) has posted “Facing Discrimination: Choosing Equality over Technology” to SSRN. Here is the abstract:

On its face, facial recognition technology poses advantages in the form of efficiency and cost-savings in sectors of society such as law enforcement, education, employment, and healthcare. However, these advantages perpetuate indirect forms of discrimination through unequal access to the technology’s benefits and—more significantly—direct forms of discrimination such as falsely identifying Black, Indigenous, and People of Color as suspects of crimes disproportionately. Facial recognition technology offers several opportunities to inject bias into its performance: through biased algorithm design, recycling racial bias in the form of past law enforcement data, and through biased user applications.

The precautionary principle warns against regulating a technology before it is fully developed and implemented, but the consequences of allowing this technology to go unregulated are overcome by the startling implications on racial discrimination in the United States. Therefore, this technology should be regulated before any further harm is done. This Comment analyzes the legislation proposed to regulate facial recognition technology by considering the longevity and breadth of the proposed regulations.