The cycle asks what these systems do to the person who lives inside them. Du Bois asks an older and harder question—the one he asked of every system he ever encountered: behind the veil of this new machinery, who is being seen, and who is being made to disappear? His frame makes visible what an individual-centred account of AI tends to miss: that the technology does not fall on everyone equally, that its judgments are most distorted and most consequential for exactly the populations he spent his life documenting, and that the decorrelation of fluency from accuracy—the cycle's description of AI's signature hazard—is not distributed at random across the people the system processes.
His concept of double consciousness illuminates a dynamic the cycle encounters when it examines how users internalise the AI gaze. When a person knows they are being watched, ranked, and scored by a system they cannot inspect, they begin to perform for the watcher. They curate their data exhaust, manage their digital footprint, and let the system's gaze become a resident editor of the self. This is double consciousness in a new register—not the white gaze of 1903 internalized into the Black soul, but the machine gaze of the present internalized into every soul exposed to it. The difference that Du Bois would insist upon is that this doubling falls more sharply on some: a facial-recognition system trained predominantly on lighter-skinned faces sees darker faces less accurately, so the scored self diverges further from the real self, and the cost of the divergence is borne by those already on the wrong side of the line.
His empirical method—the conviction that careful, honest counting is a tool of liberation, that the documented truth of a community's life is a weapon against the lies told about it—is precisely what the cycle means when it argues for AI accountability. Du Bois did not reject measurement; he out-measured the racist institutions of his day. He would not have us abandon the audit. He would have us discipline it, asking whether the categories honour the complexity of the people inside them, whether the purpose serves the measured or only the measurer.
His concept of second sight—the epistemic faculty born of necessity, the clarity that comes from being forced to understand the dominant system from inside and outside at once—is the cycle's reminder that the people most distorted by algorithmic systems are often their most acute critics. Much of the most important knowledge about what AI systems actually do has come from precisely those populations. Du Bois predicted this distribution of insight a century in advance.
William Edward Burghardt Du Bois was born in 1868 in Great Barrington, Massachusetts, and became the first African American to earn a doctorate from Harvard, completing it in 1895 with a study of the suppression of the African slave trade. In 1896 the University of Pennsylvania hired him, on insulting terms and without a proper office, to study the Black population of Philadelphia's Seventh Ward. The reigning assumption was that he would document a pathology. Instead he produced The Philadelphia Negro (1899), a work of empirical sociology so far ahead of its time that the discipline took decades to catch up: he went door to door, interviewed thousands of households, and mapped occupations, incomes, family structures, health, and housing block by block, insisting on the irreducible variety of a population the category 'Negro' had erased.
In 1903 he published The Souls of Black Folk, introducing the concepts of double consciousness and the veil that remain central to understanding race and identity. A co-founder of the NAACP and longtime editor of its magazine The Crisis, he authored the landmark revisionist history Black Reconstruction in America (1935), which reframed American history around the labour of the enslaved, and organised a series of Pan-African Congresses across decades. His thought moved steadily toward a global analysis of race, capitalism, and empire. He died in Accra, Ghana, in 1963, on the eve of the March on Washington, still asking.
The title of his masterwork captures the method: he called the book The Souls of Black Folk, not the conditions or the statistics, because the object of his science had a soul, and a sociology that captured income distributions while missing the spiritual reality of a community had failed. That dual insistence—on the rigour of the count and the irreducibility of what the count cannot hold—is his most durable gift to a moment in which measurement machines have become planetary.
Double Consciousness. The peculiar sensation, Du Bois writes, of always looking at oneself through the eyes of others—measuring one's soul by the tape of a world that looks on in amused contempt and pity. It is not merely an external indignity but an internal wound: the hostile external judgment colonises the internal one. Double consciousness maps precisely onto the condition of being processed by a discriminating data system: the individual is compressed into a proxy and the proxy is judged, and the individual must live with a scored self they cannot see or correct.
The Veil. The membrane that lets each side glimpse the other but prevents true sight, true contact, true recognition. The veil maps onto algorithmic othering in both directions simultaneously: hyper-visibility for those surveilled by predictive systems that track every movement, and invisibility for those whose data is absent or pathologised, who fall through the floor of systems that can only act on what they can measure. Du Bois located the problem in the membrane—the structure—not in the individuals on either side, which is exactly the right frame for algorithmic bias.
The Politics of Data. Du Bois's empirical method contains a lesson the age of scale-as-objectivity most needs: a dataset is a made thing, and the making is where the politics live. What gets counted, how the categories are drawn, who does the counting, and to what end—these decisions, made before a single figure is tabulated, determine what the data can possibly say. More data gathered the same way reproduces the same distortions at higher resolution. He wanted different statistics, gathered with method and conscience.
Proxy Discrimination. A century before the term existed, Du Bois documented how the colour line was enforced not only by explicit law but by a thousand indirect mechanisms, each of which encoded race without naming it. Redlining was a proxy. The poll tax was a proxy. The grandfather clause was a proxy. Discriminating data systems reconstruct the line from zip codes, names, and browsing patterns with the same efficiency while offering the alibi of mathematical objectivity.
Second Sight. The epistemic privilege of the oppressed: those forced to understand the dominant society from below and within acquire a knowledge of the whole system that the comfortable centre never needs. Second sight is not a consolation prize for injustice. It is a faculty, and its existence means that the communities most harmed by algorithmic systems are also their most accurate critics—a resource the field of AI accountability has begun, belatedly, to draw on.