Review of Race After Technology

…we might reflect upon the fact that the infrared technology of an automated soap dispenser treats certain skin tones as normative and upon the reason why this technology renders Black people invisible when they hope to be seen, while other technologies, for example facial recognition for police surveillance, make them hypervisible when they seek privacy. When we draw different technologies into the same frame, the distinction between “trivial” and “consequential” breaks down and we can begin to understand how Blackness can be both marginal and focal to tech development.¹

This book made a handful of points I found memorable:

• Even when human decision-makers are already biased, replacing them with an algorithm can be worse because any biases in the algorithm will be identical across all copies of the algorithm.
• Geographic segregation enables or worsens algorithmic discrimination. Models that are officially forbidden from seeing racial data can, for example, still learn to group people roughly by race by looking at zip code.
• Criticizing Kathryn Paige Harden (whose book I greatly enjoyed), Benjamin notes that looking to genetics for information about how to improve educational outcomes seems nonsensical when we’re not even applying the information we already have: “It is not the facts that elude us, but a fierce commitment to justice that would make us distribute resources so that all students have access to a good educational environment.”² (I don’t find the criticism of Harden, or the attempt to cast her project as a form of eugenics, convincing; my takeaway would be only that that project is relatively low-priority compared with the need to redistribute resources.)
• Sci-fi dystopias often “deflect attention away from the fact that, in the present, many people already live a version of the dystopia” and “[v]iewers … are compelled to identify with the future oppression of subordinated White people without necessarily feeling concern for the ‘old’ underclasses in our midst.”³ (I would like to believe this at least sometimes has a more positive effect than Benjamin implies; seeing a particular form of suffering afflict—in fiction—people we are already prone to empathize with, may make us more prone to recognize that form of suffering in the real world and then to extend our empathy to people we had previously overlooked.)

The message that algorithms can be very racist in very non-obvious ways is an important one. I did not feel that I gained much further insight from the book regarding causes or solutions, though. It often seems to make sweeping claims without making careful arguments for them or clearly expounding on their implications. I’ll give a couple examples.

Many tech enthusiasts wax poetic about a posthuman world and, indeed, the expansion of big data analytics, predictive algorithms, and AI, animate digital dreams of living beyond the human mind and body—even beyond human bias and racism. But posthumanist visions assume that we have all had a chance to be human. How nice it must be . . . to be so tired of living mortally that one dreams of immortality.⁴

To me, that sort of statement gives the impression of a very binary worldview, where humanity is split into two groups: the privileged who live essentially fulfilling lives, and the oppressed who do not. The human experience is not so one-dimensional; even those who are privileged in many aspects of life often still find life to be fundamentally miserable. “Posthumanist” aspirations don’t (usually, I think) arise because people have achieved a satisfactory baseline and then gotten bored; they arise because achieving that baseline in the current world is so elusive.

Another logical leap:

Racism is, let us not forget, a means to reconcile contradictions. Only a society that extolled “liberty for all” while holding millions of people in bondage requires such a powerful ideology in order to build a nation amid such a startling contradiction.⁵

Racism isn’t confined to the United States or to democratic nations, so if the argument is trying to say racism originated for that purpose, it seems unsound. Nevertheless I certainly agree with the claim that racism is frequently used to allow people to overlook contradictions in their own beliefs/behavior. But I also did not understand what role this observation was meant to play in the book—how should this knowledge change how we engage with technology or politics?