History from Below

In the AI Story

The AI discourse is overwhelmingly written from above: the CEO keynotes, the researcher announcements, the investor updates, the analyst reports. These speak in the register of decisions taken, capabilities achieved, markets captured. They construct a narrative in which the active subjects are the builders, the investors, and the policymakers, and the rest of humanity figures as users, consumers, or (in the darker framings) casualties.

History from below inverts the perspective. The active subjects become the Kenyan content moderators making sense of traumatic material at five dollars an hour; the Filipino annotators labeling training data; the Bangalore engineers discovering that their hard-won skills have been repriced overnight; the American knowledge workers whose professional identities are dissolving; the teenagers trying to understand what credentials to pursue; the parents trying to help them. The technology looks different from each of these perspectives, and each perspective contains information the top-down narrative cannot see.

The methodological challenge is the same as the one Thompson and the Annales historians faced: these populations leave fewer records than the elites, and the records they do leave have to be read carefully. For contemporary AI, this means oral histories, community organizing documents, Reddit threads, hallway conversations, the specific viral posts that reveal something the official narrative would rather obscure. Gridley's Substack post is a classic history-from-below document: an ordinary person's account of what the technology is actually doing to one household, unfiltered by the frames the industry would prefer.

The political edge is explicit. History from below does not claim that elite perspectives are false but that they are partial in ways that systematically serve elite interests. Writing history from multiple perspectives simultaneously — from above and from below — produces a more complete account than either perspective alone can provide. For AI, this means the official story and the stories of the content moderators, and the stories of the displaced, and the stories of the students trying to make sense of it all. The absence of any of these produces inadequate understanding.

Origin

The phrase became current in British Marxist historiography in the 1950s and 1960s, particularly through E. P. Thompson's work. The Annales School pursued parallel commitments from a different theoretical base. Recent applications to technology include the work of Mar Hicks, Safiya Noble, and Virginia Eubanks.

Key Ideas

Methodological inversion. The populations official narratives treat as peripheral become the central subjects of analysis.

Reading silences. Sources that elite perspectives generate systematically obscure populations whose perspectives must be reconstructed from alternative evidence.

The partiality of top-down narratives. Elite perspectives are not false but partial in ways that serve elite interests.

AI as domain. The technology's social meaning cannot be understood without systematic recovery of the perspectives the industry's official discourse excludes.

Debates & Critiques

Critics from various positions have argued that history from below risks romanticizing its subjects, substituting one partial perspective for another, or reading contemporary concerns into past populations. The defensible response is that the approach is a necessary corrective to a dominant partiality, and that its practice requires the same methodological rigor any historical approach requires.