must always sacrifice themselves for human beings, any human beings.”
“Of course,” said George Ten.
“The First Law is perhaps less satisfactory, since it is always possible to imagine a condition in which a robot must perform either Action A or Action B, the two being mutually exclusive, and where either action results in harm to human beings. The robot must therefore quickly select which action results in the least harm. To work out the positronic paths of the robot brain in such a way as to make that selection possible is not easy. If Action A results in harm to a talented young artist and B results in equivalent harm to five elderly people of no particular worth, which action should be chosen?”
“Action A,” said George Ten. “Harm to one is less than harm to five.”
“Yes, so robots have always been designed to decide. To expect robots to make judgments of fine points such as talent, intelligence, the general usefulness to society, has always seemed impractical. That would delay decision to the point where the robot is effectively immobilized. So we go by numbers. Fortunately, we might expect crises in which robots must make such decisions to be few. . . . But then that brings us to the Second Law.”
“The Law of Obedience.”
“Yes. The necessity of obedience is constant. A robot may exist for twenty years without ever having to act quickly to prevent harm to a human being, or find itself faced with the necessity of risking its own destruction. In all that time, however, it will be constantly obeying orders. . . . Whose orders?”
“Those of a human being.”
“Any human being? How do you judge a human being so as to know whether to obey or not? What is man, that thou art mindful of him, George?”
George hesitated at that.
Harriman said hurriedly, “A Biblical quotation. That doesn’t matter. I mean, must a robot follow the orders of a child; or of an idiot; or of a criminal; or of a perfectly decent intelligent man who happens to be inexpert and therefore ignorant of the undesirable consequences of his order? And if two human beings give a robot conflicting orders, which does the robot follow?”
“In two hundred years,” said George Ten, “have not these problems arisen and been solved?”
“No,” said Harriman, shaking his head violently. “We have been hampered by the very fact that our robots have been used only in specialized environments out in space, where the men who dealt with them were experts in their field. There were no children, no idiots, no criminals, no well-meaning ignoramuses present. Even so, there were occasions when damage was done by foolish or merely unthinking orders. Such damage in specialized and limited environments could be contained. On Earth, however, robots must have judgment. So those against robots maintain, and, damn it, they are right.”
“Then you must insert the capacity for judgment into the positronic brain.”
“Exactly. We have begun to reproduce JG models in which the robot can weigh every human being with regard to sex, age, social and professional position, intelligence, maturity, social responsibility and so on.”
“How would that affect the Three Laws?”
“The Third Law not at all. Even the most valuable robot must destroy himself for the sake of the most useless human being. That cannot be tampered with. The First Law is affected only where alternative actions will all do harm. The quality of the human beings involved as well as the quantity must be considered, provided there is time for such judgment and the basis for it, which will not be often. The Second Law will be most deeply modified, since every potential obedience must involve judgment. The robot will be slower to obey, except where the First Law is also involved, but it will obey more rationally.”
“But the judgments which are required are very complicated.”
“Very. The necessity of making such judgments slowed the reactions of our first couple of models to the point of paralysis. We improved matters in the later models at the cost of introducing so many pathways that the robot’s brain became far too unwieldy. In our last couple of models, however, I think we have what we want. The robot doesn’t have to make an instant judgment of the worth of a human being and the value of its orders. It begins by obeying all human beings as any ordinary robot would and then it learns. A robot grows, learns and matures. It is the equivalent of a child at first and must be under constant supervision. As it grows, however, it can, more and more, be allowed, unsupervised, into Earth’s society. Finally, it is a full member of that society.”
“Surely this answers the objections of those who oppose robots.”
“No,” said Harriman angrily. “Now they raise others. They will not accept judgments. A robot, they say, has no right to brand this person or that as inferior. By accepting the orders of A in preference to that of B, B is branded as of less consequence than A and his human rights are violated.”
“What is the answer to that?”
“There is none. I am giving up.”
“I see.”
“As far as I myself am concerned. . . . Instead, I turn to you, George.”
“To me?” George Ten’s voice remained level. There was a mild surprise in it but it did not affect him outwardly. “Why to me?”
“Because you are not a man,” said Harriman tensely. “I told you I want robots to be the partners of human beings. I want you to be mine.”
George Ten raised his hands and spread them, palms outward, in an oddly human gesture. “What can I do?”
“It seems to you, perhaps, that you can do nothing, George. You were created not long ago, and you are still a child. You were designed to be not overfull of original information—it was why I have had to explain the situation to you in such detail—in order to leave room for growth. But you will grow in mind and you will come to be able to approach the problem from a non-human standpoint. Where I see no solution, you, from your own other standpoint, may see one.”
George Ten said, “My brain is man-designed. In what way can it be non-human?”
“You are the latest of the JG models, George. Your brain is the most complicated we have yet designed, in some ways more subtly complicated than that of the old giant Machines. It is open-ended and, starting on a human basis, may—no, will—grow in any direction. Remaining always within the insurmountable boundaries of the Three Laws, you may yet become thoroughly non-human in your thinking.”
“Do I know enough about human beings to approach this problem rationally? About their history? Their psychology?”
“Of course not. But you will learn as rapidly as you can.”
“Will I have help, Mr. Harriman?”
“No. This is entirely between ourselves. No one else knows of this and you must not mention this project to any human being, either at U. S. Robots or elsewhere.”
George Ten said, “Are we doing wrong, Mr. Harriman, that you seek to keep the matter secret?”
“No. But a robot solution will not be accepted, precisely because it is robot in origin. Any suggested solution you have you will turn over to me; and if it seems valuable to me, I will present it. No one will ever know it came from you.”
“In the light of what you have said earlier,” said George Ten calmly, “this is the correct procedure. . . . When do I start?”
“Right now. I will see to it that you have all the necessary films for scanning.”
1a
Harriman sat alone. In the artificially lit interior of his office, there was no indication that it had grown dark outside. He had no real sense that three hours had passed since he had taken George Ten back to his cubicle and left him there with the first film references.
He was now merely alone with the ghost of Susan Calvin, the brilliant roboticist who had, virtually single- handed, built up the positronic robot from a massive toy to man’s most delicate and versatile instrument; so delicate and versatile that man dared not use it, out of envy and fear.
It was over a century now since she had died. The problem of the Frankenstein complex had existed in her
