There is a tsunami forming offshore. You cannot hear it yet — tsunamis are famously silent until the moment they aren't. Beneath the ordinary noise of daily life, something structural is tearing apart: the substrate on which all our institutions were built is shifting, and the institutions themselves haven't noticed. They are still operating on yesterday's clock, in a world where the clock has been replaced.
Here is what I mean by that. Every civilization runs on two kinds of time. There is the time of nature — the slow, patient accumulation of evolution, geology, and human culture, measured in generations. And there is the time of technology — the accelerating clock of innovation, measured in product cycles. For most of human history, these two clocks ran at roughly comparable speeds. The plow arrived. Farmers adapted. Centuries passed. The printing press arrived. Literacy spread. A few generations adjusted. The industrial revolution upended everything — but it took a hundred years to fully reshape society, which gave society just enough runway to stumble forward.
That runway is gone now. The gap between the two clocks — silicon speed versus biological speed — has become so wide that it is no longer a gap. It is a chasm. And we are standing on the wrong side of it.
What Silicon Speed Actually Means
People nod along when you say "AI is moving fast." Then they go back to their lives. The problem is that the human brain has no real hardware for understanding exponential change. We evolved to track predators and count fruit — linear threats, linear rewards. Exponential curves look gentle at first, then they don't, and by then it is far too late to adjust for what you misread.
Let me give you a specific example. Point an AI system at designing better chips — give it the same task it was trained to solve, but now applied to its own substrate — and it improves performance by 25% on the first attempt. A week later, 50% better. The week after that, past 100%.[2] At that rate of self-improvement, how many weeks before the system's capabilities exceed what the engineers who built it can even evaluate? We do not have a clean answer to that question. What we have is the question itself, arriving faster than our institutions are prepared to receive it.
This is not science fiction. This is the documented, peer-reviewed, publicly observable behavior of systems already deployed. And the systems being deployed today are not the ceiling. They are the floor.
"We are living through the most consequential transformation in the history of our species, and most people are too busy scrolling on their devices to notice."
What Biological Speed Actually Means
Now let us talk about the other clock. The one that governs our institutions.
A government passes a law. From first proposal to enactment, that process takes years in the best-functioning democracies, and decades in most others. Then the law is tested in courts, which operate on dockets measured in years. Precedent accumulates slowly, because the entire point of precedent is that it is stable. Regulatory agencies are staffed by civil servants who, through no fault of their own, specialize in the technology that existed when they were trained — which is to say, the technology of the past.
A university designs a curriculum. The process of approving new courses, hiring faculty, and restructuring degree programs typically takes three to five years from conception to first graduate. Which means that students enrolling today are being trained for a labor market that will exist in 2029 — a labor market that not one economist, technologist, or career counselor on earth can accurately describe. Universities that continue accepting students for careers that will not survive this decade are not being malicious. They are just running on the wrong clock.[3]
A corporation develops a five-year strategic plan. This was always a somewhat optimistic exercise, but at least it was a coherent one — the future five years from now resembled the present five years ago, more or less. That resemblance has collapsed. Five-year plans are now artifacts of a worldview that no longer applies to the world.
Governments Regulating the Wrong Thing
The most vivid illustration of the clock speed problem is watching governments try to regulate AI. They are reaching for the nearest available analogy — and the nearest available analogy is social media. Apply the social media playbook: content moderation mandates, data transparency rules, liability frameworks. It is an understandable reflex. Social media was the last big disruptive technology. The regulation worked — partially, belatedly, imperfectly — and now we have GDPR and Section 230 and a hundred half-measures that sort of address a problem that sort of resembles this one.
But AI is not social media. Social media was a distribution network for human-generated content. You could draw a clear line: human makes content, platform distributes it, harm occurs, liability attaches. AI collapses that line entirely. The system generates the content, adapts to the user, acts as an agent, and does all of this at a speed and scale that makes the concept of "review before publication" quaint to the point of comedy. Governments that try to regulate AI as if it were social media are not going to be slightly off. They are going to be categorically wrong, in ways that compound every year they persist.[4]
This is not an argument against regulation. It is an argument for regulation that understands what it is regulating — which requires building institutions that can think at something closer to silicon speed. We do not have those institutions yet. We do not even have a clear blueprint for what they would look like.
No Parachutes
In Chapter 13 of this book, I wrote about the absence of parachutes. Not a metaphor I arrived at lightly. When all hell breaks loose, not all lifelines are accessible. Some you may never count yourself lucky having access to. And the uncomfortable truth about the clock speed problem is that it offers no graceful exit for those who wait.
The industries that waited for the internet to stabilize before taking it seriously were not rewarded for their patience. They were simply gone. The newspapers that watched digital classified advertising happen slowly, then all at once, did not fail because they lacked intelligence. They failed because their internal clock — the rhythm of editorial cycles, print runs, advertiser relationships, union contracts — could not accelerate to match the external clock. The clock speed problem, once it bites, does not offer a second chance to institutions that designed themselves for a slower world.
What makes this particular moment unusual is that the thing accelerating is intelligence itself. Previous technological disruptions were powerful, but they were tools in human hands. A steam engine cannot decide to build a better steam engine. A printing press cannot redesign its own type. AI can, at least within constrained domains, and the domains are expanding. When the technology accelerating is the same technology used to accelerate technology, the feedback loop does not behave the way any prior model predicts.
Drifting Among Trajectories
I want to be honest about something. The clock speed problem does not point inevitably toward catastrophe. It points toward a choice — a choice that most of the world is currently not making deliberately. Between now and 2030, the forces reshaping society will crystallize into distinct trajectories: some regions will adapt quickly, others will not; some institutions will find ways to think faster, most will not; some individuals will thrive by updating continuously, others will freeze. The world will not choose cleanly between the best-case and worst-case scenarios. It will drift among them.
Drifting is the thing to worry about. Not the dramatic failure. Not the robot uprising. The quiet erosion that happens when systems operating at silicon speed become embedded in critical infrastructure while the humans nominally overseeing them are still running on biological time. A misaligned financial agent triggers market instability before any regulator has time to convene a working group. A logistics system fails in ways no human operator anticipated, because the humans who designed the oversight framework were imagining a slower, more legible kind of failure.[5] The danger, as I wrote in Chapter 15, is not rebellion. It is irrelevance.
"We are not ready. Not as individuals. Not as institutions. Not as a species."
What You Can Actually Do
I am not going to end this essay with a policy prescription, because I do not think policy is the primary lever available to most readers. Governments will move at their own speed, which is slow. What you do have leverage over is your own clock.
The humans who navigate this transition well will share one characteristic: they will have refused to freeze. They will have treated their own understanding as a living document rather than a credential stamped at graduation — which is precisely the insight behind Frozen Models, the previous essay in this series. Every week that you update your mental model of what AI can do is a week you are running slightly closer to silicon speed. Not all the way there. Biological speed has real limits. But closer.
The institutions will not save you from this one. They are too slow, and they know it, even if they will not say it in public. The adaptation is going to have to start at the individual level — in the way you work, learn, and think about your own relevance over the next five years. That is not a comfortable message. But comfort was always the wrong thing to optimize for when a tsunami is forming offshore.
The waves are coming. They are still silent. They will not stay that way.