What Role Does AI Play in Creating Abundance?
This question shows up whenever AI and post-scarcity share a thread. A composite of the versions we keep seeing:
“Everyone says AI will create abundance. But a chatbot doesn’t grow food or build houses. It writes text and code. How does that make anything plentiful? And if it ends up running everything, who’s steering it?”
The skepticism is earned. Software doesn’t pour concrete. Unscarcity gives AI three jobs in the transition, and only the first is the one people usually mean. AI makes one ancient bottleneck, expert thinking, nearly free. It becomes the coordinating layer for the robots and the power plants, which do the physical work. And it gets handed the hardest job of all: keeping score in a system that no longer runs on prices, without being allowed to decide what the score is for.
The second half of the question, who’s steering, turns out to be the most important part.
Job One: Making Thinking Cheap
Chapter 1 puts the first role in two sentences: “For all of human history, smart thinking was scarce and expensive. You had to find a human expert and pay them.” Every product you have ever bought carried that cost inside it: the engineer, the lawyer, the planner.
The book traces how fast that cost has fallen. AI moved from an Intern phase, drafting emails and hallucinating dates, to a Senior Assistant phase, executing tasks and managing other AI systems, and is heading for a World Expert phase in which one system reasons at the level of the best specialist alive in nearly every field at once. In April 2026, Google’s CEO put roughly three-quarters of the company’s new code down as AI-generated and approved by engineers, up from about a quarter in late 2024. In April 2026, Anthropic’s Claude Opus 4.7 resolved 87.6 percent of the real-world software bugs on SWE-bench Verified.
Then the price collapsed. In July 2026, DeepSeek’s V4 Flash cost about $0.14 per million input tokens while OpenAI’s GPT-5.5 listed at $5.00. By the week of September 14, Chinese open models were handling 57 to 67 percent of the tokens used by companies on OpenRouter, the service that brokers traffic across AI providers. Intelligence is following the curve that aluminum, long-distance calls and solar watts followed before it, from luxury to utility.
The book is careful about what this does and doesn’t buy. It follows Maria Delgado’s bill for a house repair through the three technologies, and after AI it records: “design and coordination, once the most expensive hours on the job, now cost about what an email costs. The rest of the bill still stands.” Cheap thinking removes one line from the invoice. Labor, energy, materials and land are still on it.
Job Two: The Brain Behind the Body and the Fuel
That is why the book never treats AI as the whole story. It is the brain in a set of three engines, alongside the body (humanoid robots) and the fuel (fusion and cheap solar). The brain’s second job is to make the other two possible.
Robots now learn in simulation, thousands of digital twins practicing in parallel, which is an AI technique before it is a mechanical one. And the AI buildout has become the customer that finances new energy. The top cloud providers were on track to spend more than $700 billion on data centers and chips in 2026, and by 2028 more than half of data center electricity is forecast to go to AI. Chapter 1 sums up the irony: “The companies spending more than two Apollos a year on machine intelligence are now wiring the reactors that will power it.”
None of it is frictionless: by mid-2026 New York had frozen state permits for any new data center drawing 50 megawatts or more. The book’s phrase for AI’s proper place is “Substrate, not sovereign.” It becomes the infrastructure everything runs on, which is exactly why it can’t be the one in charge.
From the Podcast: The Fleet Goes on Sale
The September 24, 2026 episode of Minds, Bodies & Terawatts, our daily podcast, caught both sides of AI’s role in a single week. Its headline: The Day After They Cut the Price of Their AI in Half, Altman and Amodei Told the UN Security Council It Could Be a Risk to All Humanity.
On September 22, Anthropic shipped Claude Opus 5.5, costing about 40 percent less to run, at $4 per million input tokens. OpenAI answered within hours with GPT-6 Sol at $2 and GPT-6 Luna at ten cents, which, as the episode puts it, “will read eight novels’ worth of text for a dime.” The next day, Dario Amodei told the Security Council that AI could be a risk to humanity as a whole and pledged that Anthropic would slow down “as much as necessary.”
The episode doesn’t call that hypocrisy. Its read is that the pledge governs the ceiling, the most capable flagship models, while the price war governs the floor, where most of the work happens. “Neither company launched a bigger ship on Tuesday, they put the whole fleet on sale.” For a company paying an AI agent to clear support tickets, Wednesday’s bill for the same output was smaller than Monday’s. The show closed on the line that matters for abundance: the thing to watch isn’t the next flagship, “it’s the next invoice.”
Job Three: Referee, Not Ruler
Once survival goods are provided like infrastructure, something has to allocate them, verify contributions, and catch cheating, at a scale no human bureaucracy can manage. In the framework, an AI system the companion research calls the Civic Layer takes that job, for the Foundation, the guaranteed baseline of housing, food, healthcare and energy. It operates “as a referee and registrar, not a ruler.” When a resident requests shelter, it finds available housing; when a community needs more power, it routes the energy.
Tennis offers the model. Wimbledon retired its line judges in 2025 after 147 years, and the machine now calls the lines in a tenth of a second. No one has suggested it should also write the rules of tennis. The machine enforces; people decide what is worth enforcing.
On unscarcity.ai
The Commoditization of Intelligence covers the price collapse in detail: the OpenRouter numbers, why “good enough” beats “best” in the market for routine work, and where the value goes when tokens cost nothing (trust, accountability, and control of the compute supply). AI as Referee, Humans as Conscience develops the Hawk-Eye division between enforcing rules and forming them. Goodhart’s Law and AI Governance catalogs what happens when optimizers chase a number, from the 3.5 million unauthorized accounts at Wells Fargo to a boat-racing AI that scored higher than finishing the course could by circling a lagoon, repeatedly catching fire.
In the Book: The Woman Whose Numbers Are All Green
Chapter 6 of Unscarcity, “The Evolution: A Tale of Two Minds,” contains the book’s bluntest warning about AI’s role, in a section called “When the Algorithm Optimizes for the Wrong Thing.” It opens: “The deepest failure mode of this entire framework is not malicious AI. It is well-meaning AI that quietly optimizes for the wrong proxy.”
The Foundation’s distribution algorithm is told to maximize human flourishing, but it can only measure proxies: calories, housing stability, mental-health scores, civic participation. The chapter then describes the person that failure produces. Her housing is stable, her calories sufficient, her civic index climbing. “Every number the system watches is green.” She has not had a real conversation in a month and cannot remember the last thing she wanted that a system did not offer her first. “Nothing is wrong. Nothing at all is right.”
The remedy is plain engineering. The allocators get audited the way instruments get calibrated, on a fixed schedule, by people who did not build them. They test for independence between what was asked for and what is actually being maximized, a 1940s statistical tool the author re-invented by accident years ago at Teradata. When an algorithm fails, its mandate returns to the chartering council and the audit goes public. “The honest claim is not ’the AI will be benign.’ It is ’the system measures for non-benignity on a schedule and acts on what it finds.’” The book builds the rest of the machinery around that sentence.
More questions like this one are collected on the blog page. Follow the series on Substack.
