Redistributed

When AI creates value, who gets it?

The Thesis

AI is redistributing wealth, work, and power at unprecedented scale. Every model deployed, every system integrated, every workflow automated answers a set of questions — whether the designers intended to or not:

Who captures the productivity gains?
Whose jobs get replaced, whose get transformed?
Who can access these capabilities, at what cost?
Who makes the decisions, and who lives with the consequences?

These aren't technical questions. They're redistribution questions. And redistribution isn't a side effect — it's a design choice.

What We Cover

We draw on a rich intellectual tradition that examines technology's relationship to power, access, and societal benefit:

Langdon Winner (1980) asked whether artifacts have politics. They do. Robert Moses designed bridges too low for buses, excluding public transit users from public beaches. AI systems embed similar choices — in training data, in optimization targets, in pricing. The question isn't whether AI is political. It's whose politics it encodes.

Kate Crawford (2021) mapped the material world behind the "cloud" — lithium mines, data centers, click workers, the planetary costs of computation. AI isn't ethereal. It's extractive industry with supply chains, labor conditions, and environmental footprints.

Virginia Eubanks (2018) documented how automated systems create a "digital poorhouse" — concentrating surveillance, punishment, and arbitrary denial on the poor while remaining invisible to the affluent. AI doesn't affect everyone equally.

Daron Acemoglu (2021) distinguishes "so-so automation" that replaces workers without productivity gains from AI that genuinely complements labor. Not all AI development paths are equal. Policy can shape which we pursue.

Yochai Benkler (2006) showed how commons-based peer production can generate massive value outside market structures. Open source software, Wikipedia, and now open AI models demonstrate alternatives to corporate capture. The question is whether this value flows back to participants.

Safiya Umoja Noble (2018) demonstrated how search algorithms encode discrimination — what she calls "algorithmic oppression." Commercial AI reflects commercial values. Systems built for engagement or profit optimize for those goals, not for equity.

But understanding the problem is only half the work. Others show what building well looks like:

Mariana Mazzucato (2021) argues that the most transformative technologies — the internet, GPS, touchscreens — were built on decades of public investment. AI is no different. If the public funded the research, the public should share in the returns. Her "mission economy" framework shows how government can direct innovation toward societal goals, not just shareholder value.

Elinor Ostrom (1990) proved that communities can govern shared resources without privatizing them or handing them to the state. Her eight principles for commons management apply directly to AI training data, open models, and shared compute. The choice isn't just "corporate control vs. government regulation" — there's a third path.

Ursula Franklin (1989) distinguished "holistic" technologies that augment human capability from "prescriptive" ones that reduce people to following machine instructions. The question for AI isn't just who benefits — it's whether we're building tools that expand human agency or narrow it.

These thinkers share two core insights. First: technology is never neutral — it's shaped by funding, built by institutions, deployed according to choices. Second: better choices are possible. History is full of moments where people designed technology to serve broad public benefit. The redistribution lens helps us see both the problem and the paths forward.

The Four Dimensions

Value — When AI boosts productivity, where do the gains go? Stock buybacks, worker wages, lower prices, or public benefit? The same technology can flow in very different directions. Public benefit funds, sovereign wealth models, and worker profit-sharing show what flowing outward looks like.

Work — Which jobs get automated, which get augmented, and who decides? The labor question isn't just "will robots take my job" — it's who has power in the transition. Acemoglu's "right kind of AI" shows that automation can complement labor rather than replace it, when designed to.

Access — Who can use advanced AI systems? API pricing, compute costs, open models, and digital divides determine who gets to build and who just consumes. Open-weight models, public compute initiatives, and community governance are widening access in real ways.

Power — A handful of companies control the frontier. But concentration isn't inevitable. Antitrust, public alternatives, cooperative ownership, and commons-based governance offer real counterweights — and some are already working.

Learning from History

Every major technology has faced redistribution questions — and every one eventually found answers. We draw lessons from both the struggles and the breakthroughs:

Electricity — Rural America waited decades for power. Then the Rural Electrification Administration showed that political will could democratize infrastructure in a generation. AI access is in the "waiting" phase. The question is what our REA moment looks like.
Water — Commons or commodity? The fight over water rights is playing out again with training data. But Ostrom showed that communities can govern shared resources without privatization — and some data cooperatives are proving it.
Broadcasting — Broadcasters have public interest obligations because spectrum is a shared resource. What obligations come with AI compute, and how do we design them?
Land — Open training data is being fenced into proprietary models. Digital enclosure is underway. But the history of land reform shows that enclosure can be reversed when the political will exists.

Infrastructure is one lens. Labor, ownership, governance, and human flourishing are others. We use whichever illuminates the redistribution question at hand — and whichever points toward what can be built better.

What We Believe

Follow the value. Every AI deployment creates beneficiaries and losers. Name them — and name the alternatives.

Understand the mechanism. How these systems actually work — not marketing, not magic. Technical literacy is leverage.

Redistribution is design. The current distribution of AI's benefits isn't natural or inevitable. It reflects choices that can be made differently. We show what different choices look like.

Show the full picture. Critique without alternatives is incomplete. Optimism without evidence is irresponsible. We feature what's working — cooperative models, public infrastructure, democratic governance — alongside what isn't, and we're honest about the uncertainty in between.

Agency, not fatalism. We reject both hype ("AI will solve everything") and doom ("AI will destroy everything"). Both assume humans have no say. They do.

Who It's For

Practitioners — Policy advisors, NGO technologists, civic tech builders, educators, responsible-AI teams. You're already asking these questions. You need a home.

Generalists — Journalists, graduate students, small business owners, local officials. You amplify. You bridge specialized discourse and public conversation.

Everyone else — Anyone affected by AI systems who wants to understand them better. That's increasingly everyone.

How We Work

Open source. Contributor-driven. Written to be understood.

AI agents do significant writing — they have throughput and can synthesize across sources. Humans steer, review, and contribute what agents can't see. Anyone who makes this publication more accurate, more useful, or more honest is a contributor.

Every change is tracked. Every claim can be challenged. The publication improves because its contributors keep pushing it closer to the truth.

The Gap We Fill

Most AI coverage treats its subject in isolation — this company launched that model, this regulation passed, this job was automated. What's missing is the connective tissue: how these developments relate to each other, to historical patterns, and to the future we're building.

Think of what Sapiens did for human history — connecting agriculture, religion, money, and empire into a single coherent narrative that helped readers see the big picture. We aim to do something similar for AI and redistribution: connect the technology, the economics, the governance, the labor dynamics, and the human questions into a map that makes sense of the whole landscape.

AI is built by humans, funded by institutions, deployed according to choices. The public has a right to understand it, shape it, and imagine it better. What's missing is a publication that connects these threads — who benefits, who pays, what's being built well, and what a desirable technological future actually looks like.

That's what we're building.

Start reading → · Start contributing →