Computing researchers who study these issues have submitted formal responses to the public call for comments regarding these new guidelines. These included a coordinated response by members of the GRAIL network, a new initiative led by the Center for Democracy and Technology (CDT) and the R Street Initiative. GRAIL’s goal is to connect technical and policy experts to inform discussions around technology policy in Washington and provide deep, rapid responses to questions of tech policy. Their response, which was led by Natasha Duarte at CDT and involved CCC Council member Suresh Venkatasubramanian, details how the different components of the new HUD guidelines are likely to make discriminatory decision making easier under the guise of easing the due process burden for landlords. Similar concerns were raised by a formal response offered by the Interdisciplinary Working Group for Algorithmic Justice at the Santa Fe Institute, which includes CCC Council member Liz Bradley, urging more transparency, both about what data these algorithms use and how they work, so that independent analysts can audit them for bias. The response also addresses what happens to liability for discrimination when these algorithms are in use.
Computing Community Consortium Blog
The goal of the Computing Community Consortium (CCC) is to catalyze the computing research community to debate longer range, more audacious research challenges; to build consensus around research visions; to evolve the most promising visions toward clearly defined initiatives; and to work with the funding organizations to move challenges and visions toward funding initiatives. The purpose of this blog is to provide a more immediate, online mechanism for dissemination of visioning concepts and community discussion/debate about them.
Responses from Computing Researchers to HUD’s Implementation of the Fair Housing Act’s Disparate Impact Standard
January 8th, 2020 / in Announcements, CCC, policy, research horizons, Research News / by Helen WrightThe following blog post is from Computing Community Consortium (CCC) Vice Chair Elizabeth Bradley (University of Colorado Boulder) and CCC Executive Council member Suresh Venkatasubramanian (University of Utah).
Algorithmic bias can be insidious, making it all but impossible to pinpoint factors that contribute to discrimination. This is particularly concerning in the context of high-stakes decisions. The new Department of Housing and Urban Development (HUD) guidelines around the use of algorithms to aid in housing decisions are an example of this. This HUD proposal acknowledges the existence of algorithmic bias but would shift much of the burden of proof to demonstrate discriminatory behavior back onto the plaintiffs, using standards for algorithmic transparency and explainability that seem unmoored from extant science about what we can hope to extract from algorithmic decision pipelines. Among other things, this would allow landlords and lenders to deflect lawsuits with an overly naive statistical approach, looking at individual factors rather than taking them in combination and thereby ignoring the potential collective effect of many lenders using the same third-party algorithm. Writing in Forbes, Elizabeth Fernandez suggests that this could undermine the Fair Housing Act.
