Tech in Policy — An Introduction
This article is a part of the Tech in Policy publication. TiP focuses on technology being used for good and shines a light on its more malicious or neglectful implementations. To read more, visit this link.
Since George Floyd was killed by police on May 25 of this year, there has begun a great reckoning in technology and society as a whole with systemic racism, bias, and how existing social and economic structures continue to perpetuate them. This is not a new conversation. Activists have been working, fighting, and dying for equality under the law and social equity as part of the modern civil rights movement since the 1950s. After all, discrimination on the basis of race and gender wasn’t broadly illegal until 1964. Gay and transgender workers weren’t protected from discrimination until barely over one month ago.
In recent decades, society has turned to technology to combat inequity on several fronts, from voting rights to healthcare, education, and beyond. This series seeks to explore how tech can improve social equity and why it needs to implement meaningful diversity to succeed.
This is the first in a series of posts that will focus on technology being used for good and shine a light on its more malicious or neglectful implementations. Though this discussion will be broad, future posts will cover specific technologies in depth.
The Census
It may seem hard to believe, but 2020 is the first year in which the decennial US Census can be mailed in, completed online or over the phone (a useful option for those with limited English proficiency), making its completion more accessible and, hopefully, more widespread. Although Graphic Information Systems (GIS) technology has been used in prior years, this is marks the first year of a partnership between a collaborative of 9 racial equity organizations and ESRI, makers of ArcGIS. Licenses for ArcGIS are being used to map hard-to-count communities of color in an effort to drive greater participation, which is crucial to determining a state’s number of representatives as well as distribution of over $675 billion in federal funds.
Voting
There are many debates underway about the current state and trajectory of American democracy, but the heart of it has always been the ability of citizens to influence local, state, and national policy through the act of voting. Unfortunately, the United States also has a long history of disenfranchising communities of color, in many cases either stripping them of their right to vote or gerrymandering districts to artfully include or exclude certain demographics at the whim of either party. This is an interesting case where technology (GIS technology again, incidentally) has mostly been used to disenfranchise voters, but has started to be reclaimed through initiatives like Pennsylvania’s Draw the Lines program, which empowers ordinary people to draw legislative maps using DistrictBuilder, an open source software for collaborative redistricting.
Education
The “digital learning gap” or “digital divide” describes a student’s lack of access to either or both high-speed internet or a computer at home to complete homework or out-of-class assignments. All the innovation in the world would still be somewhat meaningless if students have limited means to access it. A study conducted by the Pew Research Center shows that the digital learning gap exists most glaringly along racial and class lines. Where it does exist, it must be tackled on several levels. Schools have begun expanding hours beyond the end of the school day so that students can utilize school internet and computers. Municipalities have also worked to create more publicly available wifi and/or provide low income households with free wifi, particularly in the wake of Covid-19.
Projects like Personalized Learning² also seek to combine one-on-one tutoring with artificial intelligence (AI) to help bridge the digital learning gap for marginalized students.
For post-secondary school students, the Massive Open Online Course (MOOC) model has improved access to higher education. Anyone can take courses from some of the most prestigious schools in the world and either pay to take them for credit or simply gain the knowledge at no cost, a particular boon during Covid-19 that has seen enrollment in these courses spike.
Healthcare
Medical professionals are increasingly relying on telehealth to serve their patients and provide greater access to both health and reproductive care. People of color and rural communities are two groups that are impacted by a lack of proximity to affordable health care. Telehealth has begun to expand from simply conducting video appointments to incorporating artificial intelligence into patient care, enabling real-time monitoring and management of conditions such as diabetes and heart disease.
Important to note that this is a small subset of the avenues in which tech can improve racial and social equity. The potential for innovation is boundless, which is no doubt one of the most attractive aspects of working in the field.
But innovation doesn’t come without pitfalls, and the road to hell (or perhaps in this case, the road to product failure) is paved with good intentions. For every example of tech improving social equity, there is another of tech either exacerbating inequity or falling far short of its goals to improve it. In order to build products for diverse populations, companies must reflect diversity in the teams that build them through. Otherwise, they risk falling into the latter category. Consider several of the following examples of tech innovations that either benefit or disadvantage specific groups disproportionately.
Facial Recognition
The first successful facial recognition tests performed in 1967 were conducted on a database of photographs of “400 adult male caucasians” according to documents kept by the scientist, Woody Bledsoe. He and his research partner were also white men.
Even today, facial recognition faces many of the same issues that arose in Bledsoe’s first tests. The technology used by police departments often compounds one inherently biased system (arrest mugshot databases, which contain photos of Black men at disproportionate numbers) with another (facial recognition software known to misidentify Black people).
Last month, IBM, Amazon, and Microsoft halted sales of facial recognition software to US police departments, citing the need for stronger laws regulating its use. Facial recognition has long been considered a dangerous tool due to inaccuracy in identifying people of color. These misidentifications can have dire consequences, as noted by a senior policy analyst at the ACLU, “one false match can lead to missed flights, lengthy interrogations, watchlist placements, tense police encounters, false arrests, or worse.”
Speech Recognition
A similar problem has been shown to exist in speech recognition programs like Siri and Alexa. Although the applications of speech recognition are perhaps less overtly nefarious, its implications are insidious. Users who cannot be understood by these programs using their typical speech patterns are forced to either compromise their identity and assimilate to different speech patterns or not use the technology at all.
Health Tech
Although linguistic services are required for patients with limited English proficiency (LEP), software and user interfaces in the medical and pharmaceutical fields are often difficult to use for both LEP patients and the elderly. The consequences in these cases can sometimes be that patients don’t receive the treatment they need, which directly impacts their health and well-being.
Video Games, Bias in AI, Gendering Social Robots
To be clear, these are all distinct examples but I am grouping them together because they are all part of a project called Gendered Innovations, which is led by Professor Londa Scheinberg at Stanford University. It includes nearly 30 different case studies across Science, Health and Medicine, Engineering, and Environment that detail how greater inclusivity in design and development can lead to better outcomes. The whole site is worth exploring, but these are some standouts.
A quick Google search will show that plenty of players in the tech industry are pledging to improve racial equity in their companies by making stronger commitments to diversity and inclusion. Empty promises will continue to perpetuate some of the same biases already discussed and also to render equity tech much less effective. Now is the time to hold the tech industry accountable long term.
