Our firm is named after a claim about history. The claim is that societies move through long cycles, that we are now in the crisis phase of one, and that artificial intelligence is the force that will decide how this phase resolves. Fourth Turning AI Strategy is the discipline of planning for that moment: building your AI strategy on the assumption that the coming decade rewrites operating models rather than refining them.
That is a large claim. This post unpacks it in order — first the generational theory the name comes from, then the case that AI is this era's defining force, and finally what a strategy built on that premise actually contains.
The saeculum, in brief
In 1991, the historians William Strauss and Neil Howe published Generations, followed in 1997 by The Fourth Turning. Their thesis is that Anglo-American history moves in cycles of roughly 80 to 100 years — a span they call a saeculum, about the length of a long human life. Each saeculum passes through four turnings, each lasting around 20 to 25 years, and each with a distinct social mood.
A First Turning is a High: institutions are strong, consensus is broad, and individualism is muted. Postwar America from 1946 to the early 1960s is the canonical example. A Second Turning is an Awakening, when people attack those same institutions in the name of personal conscience — think of the consciousness revolution that ran from the mid-1960s into the early 1980s. A Third Turning is an Unraveling: institutions weaken, trust erodes, individualism peaks, and society fragments into camps. Strauss and Howe placed the most recent one from the mid-1980s through the 2000s.
Then comes the Fourth Turning, the crisis era. External shocks and accumulated internal decay converge, the old order proves unable to cope, and society responds not by patching its institutions but by rebuilding them. The last Fourth Turning ran from the 1929 crash through the end of the Second World War. What emerged was not a repaired version of the 1920s. It was a different machine: deposit insurance, securities regulation, Social Security, the Bretton Woods monetary system, the United Nations, the GI Bill. Institutions that had failed were not restored. They were replaced.
We should be candid about the theory's standing. Historians have criticized it as too tidy, its dating as debatable, and its predictions as flexible enough to fit many outcomes. Those criticisms have force. We hold the saeculum as a lens, not a law — and you do not need to accept every chapter of the book to find the lens useful. What matters is the pattern it points at: there are periods when institutions get patched, and periods when they get rebuilt, and the strategies that win in one period lose in the other.
Why AI lands differently in a crisis era
Two facts, taken together, make the current moment unusual.
The first is that AI is a general-purpose technology — the same economic category as steam power, electricity, and computing. General-purpose technologies do not just improve one task. They change the price of a fundamental input across the whole economy, and everything designed around the old price eventually stops making sense. With AI, the input getting cheap is a broad class of cognitive work: drafting, summarizing, classifying, analyzing, translating, first-pass reasoning. Processes built on the assumption that this work is scarce and expensive — and almost every white-collar process is — are now built on an expiring assumption.
The second fact is timing. Previous general-purpose technologies arrived into societies that broadly trusted their institutions to absorb them. This one has not. By 2023, Gallup's average measure of confidence across major American institutions had fallen to 26 percent, near the lowest level in five decades of asking the question. Operating models were already up for renegotiation before the technology arrived: remote and hybrid work unsettled by the pandemic, supply chains rewired by geopolitics, regulation in flux on three continents. AI is not landing on stable ground that it must shake. It is landing on ground that is already moving.
That combination is what the Fourth Turning frame captures. A general-purpose technology arriving in a high-trust, stable era gets absorbed slowly and politely. The same technology arriving when institutions are weak and assumptions are open becomes the organizing force of the rebuild. The firms that treat it as a feature upgrade will be working from the wrong map.
Patching versus rebuilding
There is a well-studied precedent for what happens when firms patch instead of rebuild. The economist Paul David documented it in his work on electrification. When factories first replaced their central steam engines with electric motors, most installed one large motor in the same spot and kept everything else — the overhead shafts, the leather belts, the multi-story floor plan designed to cluster machines near the power source. Productivity barely moved for roughly 30 years. The gains arrived only when a new generation of engineers redesigned the factory around what electricity made cheap: small motors at every workstation, single-story layouts, machines arranged by workflow instead of by proximity to power. The technology was necessary. The redesign was the point.
Most corporate AI adoption today is the big motor in the old factory. Assistants and copilots are bolted onto unchanged workflows, pilots multiply, and the financial results disappoint — one widely cited MIT study in 2025 found that around 95 percent of enterprise generative AI pilots showed no measurable return. The usual diagnosis is that the technology is overhyped. The electrification precedent suggests a different one: the process around the technology never changed.
Consider the pattern at a smaller scale. A 40-person logistics firm might patch by giving dispatchers a chatbot to draft customer emails. Useful, and worth doing. The rebuilding question is different: if route planning, exception handling, and customer communication can each be 70 to 80 percent automated within three years, what is the dispatch function for — and what do five experienced dispatchers do with the judgment the software cannot replicate? A professional services firm faces the same fork. Patching means associates draft documents faster. Rebuilding means confronting what happens to hourly billing when the hours collapse, and re-pricing around outcomes before a competitor does.
Patching is not wrong. It is how organizations learn, and we often recommend starting there — small, fast, and measurable. It becomes a problem only when it is the whole strategy: a portfolio of efficiency gains layered onto an operating model whose core assumptions are quietly expiring underneath it.
What a Fourth Turning AI Strategy contains
If you accept the premise — a general-purpose technology, arriving in an era when operating models are open for renegotiation — then a defensible strategy has four parts. None of them is a technology roadmap. All of them are decisions.
A candid maturity baseline
Strategy built on a flattering self-assessment fails on contact with reality. Before any thesis about the future, you need an honest account of the present: the state of your data, the skills on your team, the systems you can actually integrate with, and your organization's real appetite for change rather than its stated one. In our AI strategy engagements, this baseline is the first deliverable, and it is usually the most uncomfortable. Most organizations rate themselves a stage higher than they are. The baseline is not a grade. It is the coordinates you navigate from.
A thesis about which assumptions expire
Every industry runs on assumptions about what is scarce. Legal advice is expensive because trained judgment is scarce. Logistics margins depend on planning being hard. Agencies bill for production because production takes time. A Fourth Turning AI Strategy names the two or three assumptions your business model depends on, and states plainly which of them AI invalidates within five years. This is the hardest part to delegate, because it is a judgment about your industry, not about technology. It is also where we are most candid about uncertainty: some of these calls cannot be made with confidence yet, and a good thesis says so and names what evidence would settle them.
A portfolio of bets across horizons
A single bet is a gamble; a hundred pilots are a mess. The discipline is a deliberate portfolio across three horizons, weighted roughly 70-20-10. The first horizon is efficiency inside the current model — the patches, chosen for fast payback and learning. The second is the redesign of one function around what AI makes cheap, the way the factory was redesigned around small motors. The third is a small, genuine option on a business you could not run before. The weighting matters less than the existence of all three. A portfolio that is all first-horizon is a patching strategy wearing a strategy's clothes.
Governance that lets you move fast safely
In a crisis era, the instinct is to treat governance as brakes. We treat it as the reason you can accelerate. Clear risk tiers for AI use cases, human review where the stakes warrant it, and documented decisions about data and model use mean that when an opportunity appears, you can move in weeks instead of waiting on an ad hoc legal review — and that one bad deployment does not end the program. With the EU AI Act's high-risk obligations phasing in through 2026 and 2027, and similar frameworks following elsewhere, this is also simply the cost of operating. Our AI governance and risk practice exists because speed without guardrails is how companies bet themselves without meaning to.
Turnings reward the prepared
The Fourth Turning frame is sober, not apocalyptic. Crisis eras are not the end of anything; they are when the next order gets built, and the building is done by people who showed up with a map and a baseline while others waited for clarity that never comes on schedule. A Fourth Turning AI Strategy does not require certainty about how the decade unfolds. It requires knowing where you stand, which of your assumptions are expiring, and how you will place bets while the answers are still arriving.
If you want a candid read on where your organization stands, start with our AI Maturity Assessment — it takes about 20 minutes and gives you the baseline this post describes. And if you would rather talk it through, book a strategy call. We will be candid about what comes next.