VOLUME V, CHAPTER 14, SECTION 2
The AI Stampede, the Law's Lag and Whom It Is Written For
The previous section carried the international order all the way to the present, landing on an entirely new kind of output now arriving, artificial intelligence. This section first looks at how fast this leap is truly moving, then sets down a rule of history, that law always lags, and that once it catches up, the law it makes is, more often than not, not a barrier against vested interests, it welds the vested interest in place. Together, these two layers are the true position of law in the age of artificial intelligence.
One. This Is Another Great Industrial Leap, and Its Traits Are Speed and Reach
No great leap in human history has ever been a gentle climb, each has been a cliff-edge shift. The steam engine took decades to push people into the age of machinery. Electricity took decades. The computer and the internet took twenty years to complete the same journey. Every time, the hard-won craft of the generation before turned, almost overnight, into an exhibit in a museum. The leap now underway, the artificial intelligence leap, counting from the conversational tool that set off a global sensation at the end of 2022, has taken little more than three years, and hundreds of millions of people worldwide already use it daily, covering, in three years, a road of diffusion that once took prior generations decades to walk.
There is one further difference this time. The leaps before this mostly took the jobs of manual laborers, textile workers, factory hands, typists. This one is reaching into the professional territory of mental laborers, lawyers drafting contracts, doctors analyzing conditions, teachers preparing lessons, accountants closing books, programmers writing code, work that once demanded years of training and a license to practice, and artificial intelligence is moving quickly into all of it. What makes this interesting is that mental laborers are themselves the people who shape culture and hold the microphone, so the debate this time is louder than any before it. But loud as the debate runs, the underlying logic, capital driving technology, technology pushing out the old occupation, has not changed, and no volume of noise can block its advance, just as the Luddite movement could not block the steam engine back then. What position looks at is structure, not volume.
Two. There Is Also a Most Critical Displacement, It Has Gone Around the License Entirely
Artificial intelligence's push into mental-labor territory happens, overwhelmingly, in the civilian sphere. Military, aerospace, and defense applications of artificial intelligence are tightly governed by every government, with dedicated frameworks and approval processes. The civilian sphere runs entirely differently. An artificial intelligence tool used by a lawyer, a doctor, or an accountant is, in most cases, built by a private company, with no professional licensing threshold of any kind, and anyone can download an application and have it draft a contract, analyze a condition, or keep the books for him. The user needs no license. The maker of the tool needs no license either. Across the entire chain, the license, the very thing that has held up professional fields for centuries, has been gone around entirely. Moving at extreme speed, with extreme reach, and an extremely low threshold, and yet the territory it steps into is exactly the professional ground human society has guarded tightly, for centuries, behind the license. This displacement is the core this run will lay open further ahead.
Three. The Law's Lag Is Not an Accidental Delay, It Is a Structural Certainty
Every great industrial leap carries the same trait, technology runs out in front, society adjusts somewhere in the middle, and law and institutions fall in last of all. Western academic circles in recent years have discussed the regulatory gap around artificial intelligence a great deal, but mostly stop at describing the phenomenon, treating the law's lag as a technical problem regulators could simply close by moving faster, without seeing that it is a structural certainty. The ancestors had already carried this layer all the way through.
When the age is different, affairs are different, and when affairs are different, preparations must change. Han Fei, Han Feizi, The Five Vermin.
When the times change, affairs change with them, and the means of responding must change too, since law is always a response to something that has already happened, and always follows behind change. Laozi says, the more laws are put on display, the more thieves there are. Shang Yang says, there is no single road to governing the world, and if a thing serves the state, there is no need to cling to the old ways. Three men, from three angles, are speaking of the very same thing, that law trailing behind change is its own nature, not its flaw. What contemporary scholars call the regulatory gap is, at bottom, the modern version of exactly what these ancestors already said, only most who say it today do not know the ancestors said it first.
Four. But the Law Does Eventually Catch Up, and When It Does, Whom Is It Made For
Many people carry a simple hope, that whatever disorder a great leap causes will, in the end, be corrected by new law, that justice always arrives eventually. Lay open the legislative record of the two centuries since the first industrial system reset, and a cold, hard pattern emerges, the law that catches up is, the overwhelming majority of the time, not there to block a vested interest, it acknowledges that interest and welds it firmly in place, only once that vested interest has grown too settled to shake does it spare some scraps at the margins for the weaker side to live on.
England, after its first industrial system reset, is the clearest sample of all. The steam engine, improved by Watt from 1769 onward, spread widely, and child laborers poured into the factories by the thousands, working fourteen or fifteen hours a day. When did the law catch up. Not until 1802 was it set that a child laborer's workday could not exceed twelve hours. Not until the Factory Act of 1833 was hiring children under nine forbidden, and factory inspectors established for the first time, a mere four of them for the entire country. Not until 1847 did the Ten Hours Act arrive. From Watt's improved steam engine to an adult male worker gaining the mere right to a ten-hour day, seventy-eight years passed, and across those seventy-eight years, generation after generation of workers lived out their entire lives inside the factory. And at every single step, the factory owners fought the law fiercely, and reformers called for it again and again, and only once it was clear that without legislating, real disaster loomed, did Parliament grudgingly give an inch. The law never reached out to help on its own initiative. It was squeezed out, bit by bit.
Five. What Is Even More Worth Turning Over Is That These Late-Arriving Laws Are Often Used the Wrong Way Around
The other side of the Atlantic was the same. In the latter half of the nineteenth century, America entered its Gilded Age, and a handful of robber barons monopolized entire industries through trusts, and public resentment built for more than twenty years before the Sherman Antitrust Act finally passed in 1890, an imposing name. Yet for its first decade or so, it amounted to little more than waste paper, its wording vague, leaving enormous room for judicial interpretation, and in those years the federal Supreme Court, relying on this very antitrust law, handed down more than a dozen rulings, every one of them aimed squarely at labor unions, used to suppress strikes. A law legislated to block capital's monopoly turned, instead, into a tool used against labor. Not until 1911 was Standard Oil finally broken apart, twenty-one years from legislation to its first true bite. A senator who had pushed for the law left behind a line to the effect that the purpose of this act was never to break up the trusts, it was to calm the public's anger. Giving the public some show of an answer, while leaving capital ample room, is the most common shape law's lag takes once it finally catches up.
Shang Yang's line, turned around, pins this layer down precisely.
If it truly benefits the people, there is no need to follow the old rites. Shang Yang, Book of Lord Shang, Opening the Blocked.
Whether a law is reasonable always depends on which position the lawmaker stands in, standing with the people makes it a law that loves the people, standing with the vested interest makes it a law that welds the vested interest in place. Huang Zongxi put it even more thoroughly, that every law made after the Three Dynasties is the law of a single family, not the law of all under heaven. The law that catches up after every great industrial leap has, in the overwhelming majority of cases, matched this very verdict.
Six. Return to Artificial Intelligence
Today's discussion of law for artificial intelligence mostly runs along two directions, whether to legislate, and what shape the law should take, and which nation is moving faster than the others. The European Union passed its Artificial Intelligence Act in 2024. The United States remains deadlocked at the federal level. China issued its measures governing generative artificial intelligence services in 2023. Everywhere, an attempt at legislation is underway. But lay the two centuries of legislative history beside it, and it becomes clear, these laws will, in the end, all be made, and yet their true use, once made, is more likely to weld the vested interest in place than to block it.
The reason is not that lawmakers harbor some conspiracy, it is structural. Law always lags, and by the time it catches up, the vested interest has already taken shape, and the law either challenges it, which is exceedingly rare, or acknowledges it and spares a scrap at the margin, which is by far the most common outcome. Seeing this pattern clearly comes closer to the root of the matter than arguing over any one specific regulatory gap.
Seven. The Position Set by This Section
This section sets the forty-second layer, two layers stacked together. The first, the great leap and the law's lag is a rule of history, not a phenomenon of one time or one place, and law trailing behind change is its own nature. The second, the law made once it catches up is, in the overwhelming majority of cases, made to weld the vested interest in place, not to block it. The artificial intelligence leap moves faster than any before it, reaches wider than any before it, and has gone around the license that has guarded professions for centuries, and with these three colliding, whoever runs out first during that vacuum of no law to rely on pockets the advantage first, and by the time the law catches up, it is, more often than not, there to stamp that advantage as settled.
This is not meant to leave a reader pessimistic about artificial intelligence, it is meant to keep a reader from holding onto the simple hope that a new law will automatically arrive to see that justice is done. See this layer clearly, and every future artificial intelligence law, once it arrives, should first be asked, whose position is it welding in place, and into whose bowl does the scrap at the margin actually fall.
The next section brings artificial intelligence back to its own true position, a tool, looking at how a tool needs no license, while whoever uses the tool always does, then at where those who train the tool actually stand, and why, when something goes wrong, no one can be held to account, and finally crosses over into the volume on saving people. Carry this layer, that law lags and, once it catches up, is made for the vested interest, and walk into the next section.
Credit to the ancestors; the mistakes are mine: Tiger Lyon.