VOLUME VII, CHAPTER 1, SECTION 1
The Trap of the Tool, the Question This Volume Must Answer
The last volume, Saving People, has come all the way down its road: firefighting, culture, economy, politics, law. Five chapters now stand complete, laying down one institution after another, unlocking one lock after another. Read this far, and a question should rise up. All this unlocking along the road, taken together, what is the one whole question it was answering? This section names it straight on. That whole question is the trap of the tool. Artificial intelligence and its computing power are only a tool. They cannot replace a real human. Everything those five chapters did, saving people, unlocking, turning things back to the intent, was at root answering one face of this question. In this volume we lay it open straight on, and for the first time carry the root of saving people out to the tool itself.
One. Six Volumes, One Question
Look back at those five chapters of the last volume. On the surface each laid down its own institution and saved its own people. Yet set them side by side, and you find they answer one and the same question. The first chapter saved those who lost their living when a tool replaced them. The second chapter grew the share only a real human has, the share a tool will never hold. The third chapter made the yield large and returned it to the public, so that people need not compete with a tool at the things a tool is made of. The fourth chapter set government back in its place as service, so that a tool could not be dressed up as a coercive force standing over the people. The fifth chapter saved those who had made themselves into tools, and sealed the gap where a tool might replace the real human in holding the law.
Five chapters, five emphases, yet at root one thread. The tool's power has risen. It is replacing real humans. It is crowding out the share only a real human has. What those five chapters did was settle the people who were replaced, grow that share only a real human has, spare people the contest against a tool, keep the tool from being dressed up as coercive force, and turn those who had already made themselves into tools back into real humans. Taken together, it is one act. Put the tool back in the tool's place. Set the real human back in the real human's place. Let the tool do what a tool should do, and let the real human live as a real human should. The tool cannot replace the real human. The question this volume must answer sits inside that one line.
This thread you must trace back yourself before you can see it. Read those five chapters the first time and what you see is five separate rescues, five unlockings, five turnings back to the intent. Come to this volume and look back, and only then does it come clear all at once. Those five volumes were not five things. They were five faces of one thing, one whole question showing itself in five places. Saving people all along the road was, at root, for this: that real humans not be replaced by the tool, and that the share only a real human has not be crowded out by the things a tool is made of. This layer is the starting point of this volume, and the key for looking back at the last one.
Two. Why a Volume of Its Own
This question deserves a whole volume of its own, because it is a new question this generation has run into, one that never once came up in thousands of years.
Over thousands of years, human society met hardship of every kind. Natural disaster, plague, war, tyranny, slavery, all of these came before. Only this one never came: a tool strong enough to replace a real human. The generations before did not meet it. The thousands of years before did not meet it. This generation has run into it. Run into it, and someone has to answer it. If this generation does not answer it, the trap the next generation runs into will only be deeper, wider, harder to open. The tool's power will only grow stronger. The push to replace real humans will only grow bigger. If this generation does not make it plain that the tool cannot replace the real human, the next generation may live in a world where the real human's place has been hollowed out entirely and the things a tool is made of have taken every position. A world like that has no way back. What this volume does is this: while this generation has only just run into it, before everything has fallen, make the question plain and answer it.
Three. What Artificial Intelligence Is: However Strong, a Tool
To put the tool back in the tool's place, you first have to be clear what it actually is. In one line: artificial intelligence is the strongest tool there has ever been.
Two layers sit inside that line, and neither can be missing. The first layer: it is strong. The tools before it, the hoe, the cart, the machine, the computer, replaced the work of a person's hands and body. Artificial intelligence replaces the work of a person's mind. It can read and write, it can work accounts, it can compose writing and paint pictures, it can play chess and drive cars, it can diagnose illness, it can draft laws and judge cases, it can teach and counsel. There is almost nothing it cannot do, and it grows faster, finer, more like a real human, like to the point that you can hardly tell. This is the new hardship this generation has run into, one not seen in thousands of years. The second layer: however strong it is, it is a tool, not a person. Miss the first layer, and you cannot see its weight, cannot understand why a whole volume must be given to it. Miss the second layer, and you mistake it for something else, a god, a new species, the next civilization come to replace humankind, and so you fall into the trap where the tool takes the person's place.
Four. Strong, With a Tool's Strength
First look at its strength, how strong it gets.
It can read and write, faster, wider, more accurately than the great majority of people. It can calculate, finer, faster, steadier than anyone. It can compose writing, an article, a letter, a report, even a book. It can paint, a landscape, a portrait, a set of posters. It can play chess, Go, Chinese chess, Western chess, and beat humanity's top masters. It can drive, steadier than a person, never tired, never distracted. It can diagnose illness, reading scans, reading lab results, more accurately than some doctors. It can draft laws, drawing up one law after another to the conditions given. It can judge cases, going by the facts and the precedents to offer a proposed ruling. It can teach, mathematics, languages, programming, making clear what a person could not grasp. It can counsel, listening to a person pour out their troubles, giving advice, giving comfort. There is almost nothing it cannot do, and the more it does, the faster, the finer, the more like a real human it gets. Talk with it a little while, and you feel this is a person talking with you. Read what it writes, and you feel a person wrote it. Hear a song it sings, and you feel a person is singing. Like, to the point that you can hardly tell.
This is its strength. It is not replacing a person's hands and feet. It is replacing a person's mind. It is not helping a person do the work. It is doing the work itself. It is far stronger than any tool before it, and it is like a real human, so like that the true and the false are hard to tell apart. This is the new hardship this generation has run into, one not seen in thousands of years. Yet if this strength is where the matter stopped, a doubt would rise at once. Does this mean it is hardly different from a real human, that it can replace a real human?
Five. Like, With a Tool's Likeness
No. Not by a hair.
However strong, however fast, however fine, however like a real human, it is still a tool. A tool is a tool, not a person.
How can you tell? Look at whether, when it does a thing, there is a self of its own inside it. A person composing writing has, before the pen touches the page, a why in his heart: why am I writing this. As the pen moves, behind every word flows out his whole life's experience, his state of heart in this moment, his care for the reader, all together. When he is done he reads it back and asks, is this what I meant to say? Artificial intelligence composing writing has no why am I writing this. It takes an instruction, and from the material it was trained on it works out by probability the most likely next word, and the next after that, and pieces them into a piece of writing. When it is done it does not read it back, and does not ask, is this what I meant to say, because it has no I.
A person playing chess makes every move out of his judgment of the board, his reading of his opponent, his hunger to win, his love of the game itself, all together. Win, and he is glad. Lose, and he is troubled. After the game he goes over it again, and at home, before he sleeps, he is still turning over how a certain move should have gone. Artificial intelligence playing chess works out by its algorithm the best next move, until it wins. Win, and it is not glad. Lose, and it is not troubled. After the game there is nothing to turn over, because it has no I, no living heart that is glad when it wins and troubled when it loses. A doctor looking at a scan sees not only what is on the scan, but who this patient is, how old, how many in the family, how many more years he may live, whether he can bear the surgery and the costs that follow. When he writes down the diagnosis, he writes it carrying all of this. Artificial intelligence looking at a scan sees the data of the scan. It can pick out the lesion in it, but it has none of that share of who this patient is.
Six. Tell These Two Layers Apart, Then You Can Go On
The difference lies right here. Not in whether it can do it. It can, and faster, finer, more alike all the time. It lies in whether, when it does it, there is a self of its own inside it.
When a person does a thing, there is himself inside it, his experience, his heart, his intent, his conscience. When artificial intelligence does a thing, there is no self of its own inside it, only input, calculation, output. Its doing is a tool's doing. It gets the thing done, yet inside there is none of that share of I, my heart, my intent. So artificial intelligence, however strong, is a tool. Its strength is a tool's strength. Its doing is a tool's doing. Its likeness to a real human is a tool's likeness to a real human, and in the end it is not a real human. Between it and a hoe, a cart, a machine, a computer, there is at root no difference. It is only far stronger.
Why give a whole section to spelling out so finely that it is a tool? Because this layer is where this generation is most likely to stumble. The tools before were not like a real human. No one took a hoe for a person, a cart for a person, a computer for a person. One glance told them apart. Artificial intelligence is different. It is like a real human, like to the point that you can hardly tell. Once this generation is blinded by its likeness to a real human, and comes to believe it really is a person, really can replace a real human, really can hold a position of life and death, of giving and taking away, then the real human's place is handed over, with both hands, to a tool. With that handing over, it has fallen into the trap this volume must answer.
Both layers must be held together. It is strong: this strength must be seen, and it must be allowed to use that strength to the full in the tool's place. It is a tool: this line must be held, however like, however strong, however hard to tell apart, it is still a tool. See its strength and forget that it is a tool, and you will lift it up onto the person's place. Grant that it is a tool but make light of its strength, and you will miss the yield it could have brought about in the tool's place. Only when both layers are clear can the next step be taken steadily. The next section lays down, straight on, what a real human is, and makes plain one by one the share a tool will never hold: Keep, Guard, Love, Mercy, Heart.
Credit to the ancestors; the mistakes are mine: Tiger Lyon.
Section Two. What a Real Human Is: Keep, Guard, Love, Mercy, Heart.