In May 2014, the White House published a report on the opportunities and challenges presented by big data finding that “properly implemented, big data will become an historic driver of progress, helping our nation perpetuate the civic and economic dynamism that has long been its hallmark.” But rather than describe how to capitalize on this opportunity, the report primarily focused on how consumers, businesses, and policymakers should be wary of the potential harms of big data, based almost entirely hypothetical concerns rather than any concrete evidence. Fortunately, almost exactly two years later, the White House is starting to think about big data much more productively. Its new report, “Big Data: A Report on Algorithmic Systems, Opportunity, and Civil Rights,” recognizes that rather than being the cause of bias, discrimination, and other social and economic woes, data is a key part of the solution to these problems.
Instead of cautioning against the rise of big data and algorithmic decision-making, the report encourages its use so long as it adheres to a principle of “equal opportunity by design,” echoing a sentiment that Federal Trade Commissioner Terrell McSweeny calls “responsibility by design”—an approach that values fairness and nondiscrimination throughout the entire lifecycle of a product. To demonstrate how big data can be a tool for economic and social good, the report presents case studies for four key areas where big data is powering transformative changes: access to credit, employment, higher education, and criminal justice. For each, the report outlines the considerable opportunities for algorithmic systems to offer significant benefits, such as by analyzing nontraditional data sources to expand access to credit and powering early warning systems for police departments that can identify officers likely to use excessive force. The report also presents challenges that may arise if public or private sector actors fail to adhere to equal opportunity by design and implement these systems irresponsibly, such as by using incomplete data or failing to correctly interpret relationships between data points. Ultimately, the report correctly recognizes the potential benefits of big data and calls for increased government, academic, and private sector research to develop robust algorithmic systems that maximize the potential for this technology to be a tool for social and economic good.
There are a few likely reasons for this shift in thinking over the last few years. First, the White House hired technology industry veteran DJ Patil as the first-ever chief data scientist of the United States in February 2015, creating a channel for valuable data science expertise to better inform the White House’s approach. Second, the Obama administration has launched several promising data-driven initiatives to tackle pressing social and economic challenges in recent years that have generated enthusiastic optimism about the potential of data. For example, in the last year alone, the administration launched the Police Data Initiative to use data to improve police accountability and trust, developed the College Scorecard to use education data to help families make more informed decisions about higher education, created the Opportunity Project to use open data to make economic opportunity more accessible, and issued new rules through the Department of Housing and Urban Development to combat the lingering effects of decades-old discriminatory housing policies. Finally, this evolution in thinking is likely influenced by the fact that big data technologies, such as the Internet of Things and machine learning, are increasingly present in daily life and no longer some ominous threat on the horizon. History has shown that, from transistors to search engines, new and disruptive technologies face dramatically overinflated concerns about their potential harms that all but vanish as society is exposed to their practical benefits.
It is encouraging to see the White House begin focusing on how to maximize the benefits of big data rather than unproductively slowing these advancements by dwelling on worst-case scenarios. However, there is still a ways to go. For example, the report’s case studies provide examples of how big data could both help and harm, yet every single harm presented is hypothetical while there are multiple concrete examples of how big data is already generating social and economic benefits. Vigilantly guarding against discriminatory practices is incredibly important, yet giving hypothetical concerns equal footing with demonstrated, real-world benefits gives undue credence to the notion that big data is something to be feared. Should fear that big data will necessarily be a tool for perpetuating discrimination and undermining civil liberties dominate policymakers’ thinking, this resistance to big data will impede progress on data-driven solutions that promise to make important strides in reducing human bias in nearly every aspect of society and the economy.
Image: Cezary p.