Robert Silverberg - Enter a Soldier. Later - Enter Another

The first was a fatal change of direction that the project underwent in its early phases. The King of Spain was coming to Paris on a visit of state; and the programmers decided that in his honor they would synthesize Don Quixote for him as their initial project. Though the intelligencing program had been designed to simulate only individuals who had actually existed, there seemed no inherent reason why a fictional character as well documented as Don Quixote could not be produced instead. There was Cervantes’ lengthy novel; there was ample background data available on the milieu in which Don Quixote supposedly had lived; there was a vast library of critical analysis of the book and of the Don’s distinctive and flamboyant personality. Why should bringing Don Quixote to life out of a computer be any different from simulating Louis XIV, say, or Moliere, or Cardinal Richelieu? True, they had all existed once, and the knight of La Mancha was a mere figment; but had Cervantes not provided far more detail about Don Quixote’s mind and soul than was known of Richelieu, or Moliere, or Louis XIV?

Indeed he had. The Don—like Oedipus, like Odysseus, like Othello, like David Copperfield—had come to have a reality far more profound and tangible than that of most people who had indeed actually lived. Such characters as those had transcended their fictional origins. But not so far as the computer was concerned. It was able to produce a convincing fabrication of Don Quixote, all right—a gaunt bizarre holographic figure that had all the right mannerisms, that ranted and raved in the expectable way, that referred knowledgeably to Dulcinea and Rosinante and Mambrino’s helmet. The Spanish king was amused and impressed. But to the French the experiment was a failure. They had produced a Don Quixote who was hopelessly locked to the Spain of the late sixteenth century and to the book from which he had sprung. He had no capacity for independent life and thought—no way to perceive the world that had brought him into being, or to comment on it, or to interact with it. There was nothing new or interesting about that. Any actor could dress up in armor and put on a scraggly beard and recite snatches of Cervantes. What had come forth from the computer, after three years of work, was no more than a predictable reprocessing of what had gone into it, sterile, stale.

Which led the Centre Mondial de la Computation to its next fatal step: abandoning the whole thing. Zut! and the project was cancelled without any further attempts. No simulated Picassos, no simulated Napoleons, no Joans of Arc. The Quixote event had soured everyone and no one had the heart to proceed with the work from there. Suddenly it had the taint of failure about it, and France—like Germany, like Australia, like the Han Commercial Sphere, like Brazil, like any of the dynamic centers of the modern world, had a horror of failure. Failure was something to be left to the backward nations or the decadent ones—to the Islamic Socialist Union, say, or the Soviet People’s Republic, or to that slumbering giant, the United States of America. So the historic-personage simulation scheme was put aside.

The French thought so little of it, as a matter of fact, that after letting it lie fallow for a few years they licensed it to a bunch of Americans, who had heard about it somehow and felt it might be amusing to play with.

“You may really have done it this time,” Tanner said.

“Yes. I think we have. After all those false starts.”

Tanner nodded. How often had he come into this room with hopes high, only to see some botch, some inanity, some depressing bungle? Richardson had always had an explanation. Sherlock Holmes hadn’t worked because he was fictional: that was a necessary recheck of the French Quixote project, demonstrating that fictional characters didn’t have the right sort of reality texture to take proper advantage of the program, not enough ambiguity, not enough contradiction. King Arthur had failed for the same reason. Julius Caesar? Too far in the past, maybe: unreliable data, bordering on fiction. Moses? Ditto. Einstein? Too complex, perhaps, for the project in its present level of development: they needed more experience first. Queen Elizabeth I? George Washington? Mozart? We’re learning more each time, Richardson insisted after each failure. This isn’t black magic we’re doing, you know. We aren’t necromancers, we’re programmers, and we have to figure out how to give the program what it needs.

And now Pizarro?

“Why do you want to work with him ?” Tanner had asked, five or six months earlier. “A ruthless medieval Spanish imperialist, is what I remember from school. A bloodthirsty despoiler of a great culture. A man without morals, honor, faith—”

“You may be doing him an injustice,” said Richardson. “He’s had a bad press for centuries. And there are things about him that fascinate me.”

“Such as?”

“His drive. His courage. His absolute confidence. The other side of ruthlessness, the good side of it, is a total concentration on your task, an utter unwillingness to be stopped by any obstacle. Whether or not you approve of the things he accomplished, you have to admire a man who—”

“All right,” Tanner said, abruptly growing weary of the whole enterprise. “Do Pizarro. Whatever you want.”

The months had passed. Richardson gave him vague progress reports, nothing to arouse much hope. But now Tanner stared at the tiny strutting figure in the holotank and the conviction began to grow in him that Richardson finally had figured out how to use the simulation program as it was meant to be used.

“So you’ve actually recreated him, you think? Someone who lived—what, five hundred years ago?”

“He died in 1541,” said Richardson.

“Almost six hundred, then.”

“And he’s not like the others—not simply a recreation of a great figure out of the past who can run through a set of pre-programmed speeches. What we’ve got here, if I’m right, is an artificially generated intelligence which can think for itself in modes other than the ones its programmers think in. Which has more information available to itself, in other words, than we’ve provided it with. That would be the real accomplishment. That’s the fundamental philosophical leap that we were going for when we first got involved with this project. To use the program to give us new programs that are capable of true autonomous thought—a program that can think like Pizarro, instead of like Lew Richardson’s idea of some historian’s idea of how Pizarro might have thought.”

“Yes,” Tanner said.

“Which means we won’t just get back the expectable, the predictable. There’ll be surprises. There’s no way to learn anything, you know, except through surprises. The sudden combination of known components into something brand new. And that’s what I think we’ve managed to bring off here, at long last. Harry, it may be the biggest artificial-intelligence breakthrough ever achieved.”

Tanner pondered that. Was it so? Had they truly done it?

And if they had—

Something new and troubling was beginning to occur to him, much later in the game than it should have. Tanner stared at the holographic figure floating in the center of the tank, that fierce old man with the harsh face and the cold, cruel eyes. He thought about what sort of man he must have been—the man after whom this image had been modeled. A man who was willing to land in South America at age fifty or sixty or whatever he had been, an ignorant illiterate Spanish peasant wearing a suit of ill-fitting armor and waving a rusty sword, and set out to conquer a great empire of millions of people spreading over thousands of miles. Tanner wondered what sort of man would be capable of carrying out a thing like that. Now that man’s eyes were staring into his own and it was a struggle to meet so implacable a gaze.