VOLUME V, CHAPTER 14, SECTION 2

VOLUME V, CHAPTER 14, SECTION 2

The AI Stampede, the Law Lags, and Whom the Law It Makes Serves

The last section drew out the new output, artificial intelligence, that redraws the map. This section sets two hard rules about the law and this new output, and shows that the artificial intelligence of the present is escaping neither. The first rule is that the law always lags. The second is that when the law finally catches up, the law it makes mostly welds the vested interest in place. Call the present rush the AI Stampede, a great leap forward in the new output, everyone racing at once, and watch the two rules bite.

Start with the first rule, that the law always lags, and see why it is not an accident but a nature. Law is a made thing, written in response to what has already appeared. It cannot govern what does not yet exist, so it comes always after the thing it governs, reacting to a world already changed. In slow times this lag is small and does not much matter, the world changing at a pace the law can nearly keep. But when output leaps, when a new making races ahead at great speed, the lag opens into a gulf, and the world runs far out in front of any law meant to govern it. The AI Stampede is exactly such a leap. The new output races, remaking how things are made, decided, and known, faster than any lawmaker can draft, debate, and pass a rule, so that at every moment the real world of the new output is years ahead of whatever law exists to govern it. This is not because the lawmakers are lazy or foolish. It is the nature of law meeting the nature of a stampede. The made thing that reacts cannot outrun the making thing that races. The law lags, and in a leap it lags by a gulf.

See what fills the gulf while the law lags. In the space where the new output has raced ahead and no law yet reaches, the strongest actors move first and hold what they seize, exactly as, in a land with no law, the strong take and hold. The gulf between the racing output and the lagging law is a lawless space, and lawless space is filled by the strong, who plant themselves in the new territory, capture its heights, and dig in, before any rule arrives to govern the taking. By the time the law catches up, the new ground is already held. This is the first rule's sting. The lag is not neutral time in which nothing happens. It is precisely the window in which the new output's scarce heights are seized by whoever is strongest and fastest, unchecked, so that the law, when it finally arrives, arrives to a territory already occupied.

Now the second rule, that when the law catches up, the law it makes mostly welds the vested interest in place. This is the deeper and harder rule, and it follows from the first joined to the whole volume. When the law finally arrives at the new output, who shapes it. The vested party of the new output, the ones who seized its heights during the lag, are by then the strongest actors in the field, with the most at stake and the most weight to bring, and they reach the pen exactly as the vested party always reaches the pen. So the law that catches up is written, in large part, under the influence of those who already hold the new ground, and it tends to write their holding into law, to weld the vested interest in place. The rule that arrives to govern the new output often does not pry the heights back open. It fixes the seizure that happened during the lag, converting the fastest actors' grab into lawful, defended possession. The law lagged while the strong seized, and then the law, catching up, made the seizure legal. Both rules together are brutal, and the AI Stampede is running straight through them.

Trace it plainly in the present rush. During the lag, the actors with the most resources race ahead, capturing the new output's commanding positions, the models, the data, the infrastructure, the talent, seizing heights that were briefly open and are now held. And as the law begins, at last, to reach toward artificial intelligence, the shaping of that law draws heavily on exactly those actors, the ones who hold the heights, because they hold the expertise, the weight, and the stake. So the early law of the new output tilts, as the volume would predict, toward welding in place the positions seized during the lag. This is not a claim that every rule is corrupt or every lawmaker captured. It is the plain working of the two rules, the lag that let the strong seize, and the catching-up that fixes the seizure, playing out in the present as they have played out in every earlier leap of output.

Here the volume's discipline and its honesty both hold. To name these two rules is not to say the law should not try to govern the new output, or that all regulation is a fraud, or that the lag can be wished away. Law must try to catch up, and some law is far better than none, and a rule that even partly checks the seizure is worth having. It is to see clearly what the law is up against and what it tends to become, so that a person is not naive about it. The reader who knows the two rules watches the emerging law of artificial intelligence with open eyes, asking of each new rule the volume's standing question, does this pry the new output's heights back toward the many, or does it weld the seizure of the lag in place. Neither answer is guaranteed. The rules describe the tendency, not an iron fate, and the tendency can be resisted by exactly the forces the volume named throughout, a check over power's head, the pushing of those shut out, the rare lawmaker who holds the measure. But the tendency is strong, and a person who does not see it will mistake the welding for justice.

And here the volume's spine shows once more. The two rules are not new to artificial intelligence. They are the old pattern, the strong seizing the scarce thing and the law fixing their hold, appearing in the newest output. The AI Stampede is the line of land and the eight modern tools, running again on new ground, at higher speed. What land was across three thousand years, what the patent and the credential and the rest are in the present, the commanding heights of artificial intelligence are becoming now, a new scarce thing seized by the strong during the law's lag and welded in place by the law's catching up. The reader who walked the earlier chapters is not seeing something wholly strange here. He is seeing the volume's oldest pattern in its newest and fastest form, which is exactly why the volume can end on it, because it shows that the root the whole book dug to does not change when the output changes. Only the ground it plays out on is new.

So this section sets a forty-second layer. The law always lags, by its nature as a made thing that reacts, and in a leap of output the lag opens into a gulf that the strongest fill by seizing the new ground unchecked. And when the law catches up, it mostly welds the vested interest in place, because the ones who seized the heights during the lag reach the pen and write their holding into law. The AI Stampede is running through both rules now, the strong seizing the commanding heights of the new output during the lag, and the emerging law tending to fix that seizure. This is not a call to abandon the law but to watch it with open eyes, asking of each new rule whether it pries the heights open or welds the seizure in place, knowing the tendency is strong but not fated. It is the volume's oldest pattern, the strong seizing the scarce thing and the law fixing their hold, in its newest and fastest form. The next section goes to the core of the new output and the law, that artificial intelligence is a tool and not a subject, that the licensed are always people while the ones who train it mostly hold no license, and that the replaced stand on an ever-narrowing ground, and it ends on a paradox set out and left unresolved. Let the reader carry these two rules into the section that follows.

Credit to the ancestors; the mistakes are mine: Tiger Lyon.