Capítulo 1
When Technology Reinforces Racism: The Invisible Codes That Shape Our World
Imagine a beauty contest where thousands of contestants from over 100 countries submit their selfies to be judged not by humans but by artificial intelligence. The algorithm would select winners based purely on objective standards of beauty, free from human bias and prejudice. When Beauty AI ran exactly this contest in 2016, the results were shocking-nearly all winners were white. Despite developers' dismay, this wasn't a glitch but a revelation of how machines learn to reproduce our existing prejudices. This is the world of the "New Jim Code" that Ruha Benjamin explores-technologies that appear neutral or even beneficial while encoding and reproducing racial hierarchies. In an era where algorithms increasingly determine who gets jobs, loans, healthcare, and freedom, Benjamin's analysis couldn't be more urgent. Embraced by tech critics and social justice advocates alike, this book has become required reading in Silicon Valley companies grappling with the unintended consequences of their innovations.
Capítulo 2
The New Jim Code: How Discrimination Gets Digitized
Growing up in Los Angeles, Ruha Benjamin experienced firsthand what it meant to live under constant surveillance-from police pat-downs at playgrounds to helicopter monitoring that shook her family's roof at night. These experiences weren't isolated incidents but part of a systematic pattern of surveillance that disproportionately affected communities of color. Today, this surveillance continues in the digital realm, often without the audible rumble of helicopters but with equally devastating effects. The New Jim Code refers to new technologies that reflect and reproduce existing inequities while being promoted and marketed as more objective than discriminatory systems of the past.
Consider how even names function as social codes that open and shut doors in our increasingly digital world. Research shows job seekers with white-sounding names receive 50% more callbacks than those with Black-sounding names-an advantage equivalent to eight additional years of work experience. Even algorithms trained on human writing associate white names with "pleasant" words and Black names with "unpleasant" ones. This bias extends to automated hiring systems, credit scoring algorithms, and facial recognition software, which consistently show lower accuracy rates for people of color.
This exemplifies how technical systems encode human judgments and societal biases. When Facebook's algorithms reproduce racist patterns from past users' behavior, or when Google's search results reinforce stereotypes about gender and race, they're not neutral tools but active participants in perpetuating discrimination. Unlike Michelle Alexander's concept of "The New Jim Crow," which described how the carceral system creates a racial caste system through colorblind ideology, the New Jim Code embeds cultural coding into technical coding, making discrimination both more pervasive and harder to detect.
What makes this particularly insidious is that tech designers encode judgments into systems while claiming racist outcomes are external to their process. This creates a double layer of racism-magnified and buried under digital denial. Companies often defend their algorithms as "race-blind" while failing to acknowledge how historical biases in training data perpetuate discrimination. The defining characteristic of these technologies isn't just that they discriminate, but that they do so while claiming objectivity or even benevolence, often under the guise of efficiency and innovation.
Why does this matter? Because Silicon Valley's "Move Fast and Break Things" ethos raises the critical question: what about the people broken in the process? Real-world examples abound: when a woman with a tumor is denied a bank loan despite high income and no debt because an algorithm flagged her as high risk, we see how "data sharing" allows stigmas to follow people efficiently across systems. Similar cases include facial recognition systems failing to identify people of color, healthcare algorithms underestimating the care needs of Black patients, and predictive policing software disproportionately targeting minority neighborhoods. As computer scientist Donald Knuth warns, "algorithms are getting too prominent in the world," and their unchecked deployment risks automating and amplifying existing social inequities at an unprecedented scale.
Capítulo 3
Engineered Inequity: When Discrimination Is By Design
Can robots be racist? They certainly can, even without explicit intention to harm. The assumption that racism requires "white hoods and racial slurs" fundamentally misunderstands how bias operates in our society and our technology. When machines become more "intelligent" by learning to think like humans, they risk becoming more racist as they absorb our cultural biases, often in ways more subtle yet equally harmful as human prejudice.
Machine learning algorithms learn from datasets that reflect deeply ingrained cultural prejudices. For example, facial recognition systems consistently show higher error rates for darker-skinned faces, particularly women of color, due to training data skewed toward white male faces. Resume screening algorithms have been found to penalize candidates with names typically associated with minority communities, perpetuating workplace discrimination. Even with diverse programming teams, these systems can perpetuate stereotypes if developers lack the institutional power to intervene when problems arise or if economic pressures prioritize speed over equity.
The term "stereotype" provides a perfect entry point for understanding technology's default settings. Originally referring to printing plates making copies, it evolved to mean "image perpetuated without change" before becoming shorthand for beliefs about groups. This etymology reveals crucial connections between technical and social systems, showing how mechanical reproduction of bias mirrors societal patterns.
Seemingly trivial examples reveal deeper systemic issues. Soap dispensers that don't detect darker skin treat certain skin tones as normative, while medical devices like pulse oximeters show less accuracy with darker skin tones. The same Black people rendered invisible by these everyday technologies become hypervisible through police surveillance technologies, facial recognition systems, and predictive policing algorithms. These connections reveal how Blackness is simultaneously marginal and focal to tech development, either ignored or over-scrutinized.
China's social credit system exemplifies explicit ranking with far-reaching consequences. Using proprietary algorithms, it tracks not just financial history but education, work history, purchases, and leisure activities in unprecedented detail. Playing video games marks someone as "idle" while buying diapers suggests responsibility. Low scores bring concrete penalties like restricted travel and limited internet speeds, while high scores offer privileges like faster service, preferential loan rates, and VIP access to services.
Western companies already collect similar data through credit scores, loyalty programs, and social media profiles. China's distinction is consolidating these behaviors into a comprehensive "citizen score" that determines access to jobs, mortgages, and transportation. The system's architects claim it "will ensure bad people don't have a place to go, while good people can move freely." Even one's social network affects their score - associating with low-scoring individuals can damage one's own rating. This creates digital castes and reinforces existing social hierarchies, demonstrating how algorithmic systems can formalize and amplify societal inequities through seemingly objective metrics.
Capítulo 4
Default Discrimination: When Neutrality Reproduces Inequality
What appears as technological "glitches" often reveals deeply embedded systemic racism. When Google Maps mispronounces "Malcolm X Boulevard" as "Malcolm Ten Boulevard," or when voice recognition systems struggle with non-Western names and accents, these aren't mere technical oversights. Rather, they reflect the cultural assumptions and blind spots of a predominantly white, male programming workforce whose worldviews become invisibly encoded into systems that shape daily navigation, communication, and decision-making for millions.
Risk assessment algorithms in the criminal justice system demonstrate how seemingly objective tools can perpetuate racial inequities while maintaining a veneer of neutrality. These systems process variables that are already structured by historical racism - including criminal history, education level, employment status, and zip code - without explicitly considering race. ProPublica's landmark investigation revealed stark racial disparities, with Black defendants being falsely labeled high-risk at nearly twice the rate of white defendants (45% vs 23%). Even more troubling, white defendants labeled low-risk went on to commit new crimes at higher rates than predicted, showing how the algorithms systematically underestimated their risk while overestimating that of Black defendants.
Predictive policing software like PredPol, which adapts earthquake aftershock models to crime prediction, exemplifies how algorithmic systems can create dangerous feedback loops. By designating "temporary crime zones" based on historical data, these tools direct increased police presence to neighborhoods that have historically experienced over-policing due to racial bias. This heightened surveillance inevitably leads to more arrests for minor infractions, generating new data points that justify continued police concentration in these areas. What begins as crime prediction transforms into crime production through a self-reinforcing cycle of surveillance and enforcement.
Using The Matrix as an analytical lens, these technological "glitches" expose the underlying architecture of social inequality. Just as the deja vu cat in the film signals fundamental problems with the system's code rather than isolated errors, racist technological failures reveal how deeply social biases are embedded in our technical infrastructure. The glitches make visible what is usually invisible: the constructed nature of our social reality and its encoding in digital systems.
Google search results provide an instructive example of how algorithmic bias can be addressed when there is institutional will to do so. Historical searches for "Black girls" predominantly yielded pornographic content, reflecting and reinforcing racist hypersexualization. After sustained criticism and activism, Google implemented changes resulting in more diverse and empowering results, including organizations like Black Girls Code. This transformation demonstrates that the technical capability to mitigate bias existed all along - what was missing was social awareness and corporate motivation to implement solutions.
Other persistent examples of algorithmic discrimination continue to surface: image searches for "unprofessional hairstyles" disproportionately show Black women while "professional hairstyles" display white women; job recruitment algorithms show high-income positions to men more frequently than women; and dynamic pricing systems charge higher rates for SAT tutoring in predominantly Asian neighborhoods. As artificial intelligence systems gain greater agency in society, these embedded cultural biases become increasingly problematic. The danger lies not just in individual instances of discrimination, but in how these technologies claim to transcend human prejudice while actually encoding existing social hierarchies in more subtle, technically legitimized forms.
Capítulo 5
Coded Exposure: The Paradox of Visibility and Vulnerability
Technology alternately renders Black people invisible or hypervisible in harmful ways. The sitcom "Better Off Ted" brilliantly satirizes this through an episode where motion sensors fail to detect Black employees, preventing them from using elevators, doors, and water fountains. Rather than fixing the technology, the company installs separate manually operated facilities-a clear parallel to Jim Crow segregation-before eventually hiring white employees to follow Black workers so sensors will activate. This absurdist scenario reflects very real technological biases that continue to plague modern systems.
This dual nature of technological exposure for Black people manifests in countless ways: automated soap dispensers that don't recognize dark skin, security cameras that disproportionately flag Black individuals as suspicious, and facial recognition systems that simultaneously fail to identify Black faces while over-surveilling Black communities. This paradox reflects broader power dynamics where technology either erases or over-polices Blackness. Exposure operates as both a technical process (like exposing film) and a form of vulnerability, interweaving artistic, scientific, and sociopolitical dimensions.
The historical relationship between visual technologies and racial taxonomies runs deep. Victorian-era photography emerged alongside pseudo-scientific theories of race, with early photographers explicitly using their medium to document and classify human difference in service of colonial projects. These images helped construct and reinforce racial hierarchies while claiming scientific objectivity. Even modern color photography, despite its claims to neutral physics, embeds bias through choices about light sensitivity, color balance, and exposure settings that privilege lighter skin tones.
Kodak's Shirley Cards, used from the 1950s through the 1990s, epitomize how technical standards encode racial bias. These calibration cards exclusively featured white women, establishing whiteness as the technical default and systematically underexposing darker skin tones. Black subjects appeared blurry, featureless, or invisible in photographs. Three major factors finally forced change: the practical needs of school integration photography, corporate demands for accurate product photography of brown goods, and economic pressure from Asian markets where light-skin bias limited market share.
The Polaroid ID2 camera case study reveals technology's direct role in systems of racial oppression. The camera's "boost button" for darker skin wasn't an inclusive feature but rather a tool for South Africa's apartheid regime to photograph Black citizens for mandatory passbooks. The Polaroid Revolutionary Workers' Movement's powerful slogan - "Polaroid Imprisons Black People in 60 Seconds" - sparked a seven-year campaign that ultimately pushed Polaroid to exit South Africa, creating a model for broader corporate divestment movements.
Contemporary examples show these patterns persisting in digital technologies. HP faced criticism when their webcam tracking software failed to recognize dark-skinned users, citing technical limitations rather than examining underlying biases. Google Photos' algorithmic labeling of Black people as "gorillas" revealed how historical racist taxonomies continue to shape modern systems. While surveillance technology affects everyone, its impacts remain deeply unequal - facial recognition systems demonstrate both technical failure to accurately identify Black faces and discriminatory deployment targeting Black communities through law enforcement databases. These ongoing issues demonstrate how technological exposure continues to operate as a mechanism of racial control and discrimination.
Capítulo 6
Technological Benevolence: When "Help" Perpetuates Harm
What does it mean to "fix" something? To repair, to get even with, to make stable, or to influence through improper methods? This question lies at the heart of technological benevolence-technologies marketed as solutions to social problems that often extend rather than disrupt systems of oppression. The very language of "fixing" reveals underlying assumptions about what constitutes a problem and who gets to determine the solution.
Electronic monitoring (EM) exemplifies this complex dynamic. Marketed as a humane alternative to mass incarceration, these devices create what researcher James Kilgore terms "e-carceration"-a digital prison without walls. During his own experience with EM, Kilgore faced impossible choices: taking his seriously ill mother to the hospital meant violating his mobility restrictions. The technology's rigid parameters failed to account for basic human needs and family obligations. Similar stories abound: parents missing children's school events, workers losing jobs due to device malfunctions, and individuals facing re-arrest for technical violations like low battery alerts.
The pre-trial implementation of EM reveals deeper systemic issues. Predominantly used for those who cannot afford bail, it creates a two-tiered justice system where wealth determines freedom. The algorithmic risk assessment tools determining monitoring assignments frequently incorporate historical arrest data and zip codes-factors heavily influenced by racial profiling and segregation. Studies show these algorithms consistently rate Black defendants as higher risk, despite equivalent criminal histories. One analysis found monitoring rates in predominantly Black neighborhoods were three times higher than in white areas with similar crime statistics.
AI-powered hiring tools represent another troubling iteration of technological benevolence. Platforms like HireVue claim to "reduce unconscious bias" through facial expression analysis and comparison to "successful" employees. While one healthcare company celebrated a 79% reduction in hiring time, candidates report dystopian experiences-performing for algorithms without feedback or recourse. Amazon's abandoned AI recruitment tool, which systematically downgraded resumes containing words like "women's" or graduates of all-women colleges, demonstrated how historical hiring biases become encoded in seemingly neutral systems.
Medical technologies particularly illustrate how racial assumptions become embedded in scientific instruments. Spirometers, used to measure lung function, include "race correction" buttons that assume lower capacity for Black patients-a practice traced to plantation physician Samuel Cartwright's racist theories. This standardized difference proved devastating during a 1999 class-action lawsuit when Black asbestos workers struggled to prove injury because their "normal" baseline was artificially lowered. Similar race-based adjustments persist in kidney function tests and cardiac algorithms, perpetuating health disparities under the guise of scientific precision.
Healthcare "hot-spotting" exemplifies how even well-intentioned data use can reinforce stigma. This approach, borrowed from police profiling techniques, maps high-cost patients to target interventions. While identifying care gaps, it often characterizes minority communities as inherently problematic, "socially disintegrated," and "unable to self-care." One study found that hot-spot designations correlated more strongly with race than with actual health outcomes. Such technological "solutions" offer surface-level fixes while leaving structural inequities intact, becoming what scholar Ruha Benjamin calls "permanent placeholders for bolder change."
Capítulo 7
Beyond Empathy: Rethinking Technology's Role in Social Change
The growing "empathy" discourse in technology, particularly virtual reality marketed as an "empathy machine," raises critical questions about the authenticity and effectiveness of technologically mediated empathy. While VR experiences claim to help users understand others' perspectives, they often reduce complex social issues to simplified narratives. Mark Zuckerberg's tone-deaf virtual "visit" to hurricane-devastated Puerto Rico, where his avatar smiled while floating through flood scenes, exemplifies how these technologies can trivialize real suffering. Rather than fostering genuine understanding, many VR experiences create "poverty porn" that reinforces existing power dynamics between privileged viewers and marginalized subjects.
The limitations of technological solutions become particularly apparent in criminal justice applications. While VR training programs for prisoners promise to aid rehabilitation and job readiness, they collide with the harsh reality that over 65 million Americans face severe employment discrimination due to criminal background checks. Companies may invest in VR training while maintaining hiring policies that exclude former offenders. Paradoxically, VR is increasingly deployed to generate empathy for law enforcement officers, while similar efforts to understand the experiences of those affected by police violence remain relatively uncommon.
Design thinking methodology, despite its popularity, requires careful scrutiny. The concept of "human-centered design" often prioritizes the needs of paying clients or dominant groups rather than those most impacted by the solutions. While the Design Justice movement attempts to correct this by promoting marginalized communities' participation in technology design processes, deeper questions persist about design thinking itself. The framework's emphasis on rapid iteration and constant forward momentum can preclude necessary reflection and consideration of historical context.
The tech industry's approach to diversity often focuses on teaching marginalized individuals to code-switch - adapting their behavior to conform to dominant cultural norms. This places the burden of change on already disadvantaged groups rather than addressing systemic inequities. A more transformative approach would focus on rewriting dominant cultural codes and embedding new values into our social relations and technological systems. This might include alternative success metrics beyond efficiency and profit, such as community well-being and social equity.
Creating more just technologies requires fundamental shifts in priorities and processes. This means moving beyond surface-level solutions that maintain existing power structures toward approaches that prioritize equity over efficiency, social good over market imperatives, and genuine transformation over technological quick fixes. It requires questioning not just how we design technology, but who benefits from it and whose knowledge and experiences are valued in its creation.
Capítulo 8
Abolitionist Tools: Reimagining Technology for Liberation
What would it mean to build truly liberatory technologies? Benjamin advocates for coded equity audits as essential tools for accountability in tech. Auditing-traditionally used in finance but also applied to expose discrimination in housing and employment-can be adapted for algorithmic systems. Such audits must be independent and enforceable, not just internal corporate processes. These audits should examine not only outcomes but also the underlying assumptions, training data, and decision-making processes that shape algorithmic systems.
Individual-level data rights like those in Europe's GDPR often exempt law enforcement and can deepen existing inequities. While GDPR provides important protections, its focus on individual consent and control fails to address systemic power imbalances. Community-based approaches like the Detroit Digital Justice Coalition's "DiscoTech" model and the Our Data Bodies project center marginalized perspectives and offer more promising directions. These initiatives emphasize collective data governance, participatory design, and community ownership of technological infrastructure.
Benjamin connects current data justice work to historical Black resistance through data, citing W.E.B. Du Bois's pioneering infographics and statistical work, and Ida B. Wells-Barnett's meticulous documentation of lynching statistics. These historical examples demonstrate how data collection and visualization can serve as powerful tools for social justice and resistance. Benjamin emphasizes that narrative tools are as essential as computational ones in challenging the master narrative that technology is neutral.
Moving beyond mere access or charity, Benjamin advocates for solidarity-based approaches that question who defines, distributes, and hoards "fortune" and how it was obtained. This requires examining the political economy of technology development and deployment, including venture capital funding patterns, corporate ownership structures, and labor practices. She highlights Virginia Eubanks' warning from welfare recipients tracked through EBT cards: "You should pay attention to what happens to us. You're next," illustrating how surveillance technologies often target vulnerable populations first before expanding to broader society.
Against the overwhelming whiteness of popular future visions, Benjamin points to Afrofuturist and Chicanofuturist alternatives that center people of color while confronting systems of domination. These alternative futures imagine technologies that enhance community autonomy, cultural preservation, and collective liberation. She showcases subversive projects like the "White-Collar Crime Risk Zones" app that flips predictive policing to target financial crimes, and Hyphen-Labs' designs like police-recording earrings and facial recognition-blocking clothing. These projects demonstrate how technology can be repurposed to challenge rather than reinforce existing power structures.
For Benjamin, liberation requires both critical analysis and creative alternatives that bring joyful, emancipatory technologies to life. This means developing technologies that prioritize community needs, foster collective agency, and challenge oppressive systems. We must disrupt the master narrative that technology is neutral and loyal to those in power. Emancipatory designs aren't just theoretical-they already exist in community technology centers, cooperative platforms, and grassroots innovation spaces that point toward possibilities for technology that serves justice rather than reinforcing oppression.
What would our world look like if we designed technologies with justice at their core? How might we code for liberation rather than control? These questions demand practical exploration through participatory design processes, community-led technology assessment, and democratic governance of technological systems. The New Jim Code isn't inevitable; it's a choice we make with every line of code, every design decision, every implementation. Another world is possible, and it begins with reimagining the relationship between technology and justice through concrete practices of resistance, creation, and collective liberation.
