Whatever the bureaucrats might have decided, Jeffrey Praxis could see that reconstruction of this mammoth mess would be a growth industry for the next couple of years. So he went to San Francisco, bent the knee to his grandfather, pledged fealty to his wicked aunt, and swore peace with his cousins Brandon and Paul, who had fought in the war for the other side. In fact, swapping war stories—and quietly trying to figure out if any of them had ever gotten the other in their sights—had been a major part of their bonding process.
Jeffrey became rich along with the family company and soon was a project manager and member of the executive team. When the time came for PE&C to apply for the Bi-Cities Rail Transit job, he was a natural to take charge. Except … by that time most of the work was done by computer intelligences, with human beings like him standing around watching, making the occasional policy and money decisions, and asking insightful questions.
Now, as he rode the cab of MOLE 2, Jeffrey sensed the need for one of those questions. When Moley mentioned “the end,” he was talking about the end of the line, when his cutter head had completed the tube under Los Angeles. So it was time for the human member of the team to tread delicately.
“You still have a long way to go,” Jeffrey said. “Through some difficult ground, too.”
“I know that,” the intelligence said. “I calculate twenty-three more head changes before the end.” That was the process where MOLE 2 backed away from the rock face by about a meter and directed a team of ’bots in exchanging worn carbide teeth and planing surfaces for new ones. “And after that …”
Jeffrey wondered what an artificial intelligence’s conception of death might be. Moley had to know that a rig assembled and operating so far underground would never see the light of day. Once it reached its terminus in Los Angeles, Moley would be instructed to drive a hundred meters further on and fifteen meters further down, stop digging, and set the batch team to backfilling the rig’s working cavities and the tunnel space behind it. They would create a plug of solid ground where before there had been a working machine.
“After that,” Jeffrey said, trying to sound jovial, “we’ll find you, your brothers, and your ’bot teams another project to run. Maybe a steam shovel in a nice open-pit mine, someplace above ground, in the sunlight.”
“I was coded for this tunneling project,” Moley said somberly. “Its parameters are part of my essential data structures.”
“But you can learn, can’t you?”
“I have projective capacity.”
“Well then, there you go.”
* * *
“Yeow!” Jacquie Wildmon retained just enough sensory awareness to find the cutoff switch to her cortical leads and just enough motor control in her left hand—the side of her brain opposite the amped inputs—to flick it. The blinding swarm of kinetic images, nonverbal facts, random symbols, jumbled words, numbers with impossible decimal places, and blurred data streams died inside her head.
“That hurt!” she said aloud to the empty laboratory.
“Do you require medical assistance?” Vernier asked.
Intelligences didn’t need names, of course, just access numbers for protocols and routing. But humans usually needed names for addressing the machines verbally. Jacquie had chosen this name because, like the sliding device on a graduated scale or caliper, the intelligence had a special subroutine that allowed it to throttle its own cycle speed at will. Vernier was a test engine for a new kind of interface.
“No, just a minute to sort out my thoughts,” Jacquie said. And after that minute, “Maybe a couple of aspirin, too.”
A housekeeping ’bot—an entity shaped like a scorpion with corrugated rubber on its pincers, sticky pads on the tips of its eight legs, and a prehensile tail that ended in a data-and-charging plug—scrambled across the floor, climbed the wall under the first-aid kit, opened it, and retrieved a bottle of pain pills. It brought her the bottle but forgot a glass of water to wash the pills down with. AIs didn’t have throats, didn’t need to swallow, and so sometimes forgot the mechanics of the human body if that knowledge wasn’t part of their programming.
“Thanks,” she said anyway and popped two of the pills dry.
Jacquie had worked at Tallyman Systems ever since her father, Richard Praxis, brought her into the company, which was as soon as she got her degree in mechanical engineering. That was during the war, and she had stayed on after the peace because she found the subject of computer-mechanical interfaces—loosely, “robotics”—interesting. And then, as the computing half moved into the realms of autonomy, self-direction, motivation, and human emulation, she graduated from mechanics to heuristics, or the science of problem solving, discovery, and learning based on past experience instead of programmed formulae.
Early on, she had consulted on the periphery of Tallyman’s experiments with stochastic-evolutionary programming. According to rumor, that department had contributed greatly to ending the war. From her own personal experience, she also knew that the department’s work had later forged the original link between Tallyman and her father’s old company, Praxis Engineering. That link had nearly come apart when her father died, for reasons that were still a mystery to her.
But then the link had been made stronger—and had made both companies rich—by the big earthquake up in San Francisco. In its wake, rebuilding the water, sewage, transit, and electrical grids in nine counties had relied heavily on her company’s proprietary Stochastic Design and Development® software package. But at the same time, the earthquake had been a net setback for Jacquie’s own field. The disruption created a major recession in the country and created an instant demand for thousands of human hands to dig out, rebuild, and reorient the damaged area. Nobody much wanted to hear about how to give this work to teams of ’bots guided by primitive intelligences that needed hours of teaching and training to learn how to pick up a brick here and put it over there.
Without the quake, the domain of artificial intelligence might have grown slowly and opportunistically, built stature for itself, and expanded into the surrounding society—like any new, evolving species. Instead, the whole field was put on pause, strangled for funding, and left almost stillborn in the laboratory—except for a few diehards like Jacquie Wildmon. Tallyman and its competitors in heuristics had to wait years, until the economy recovered. Then growing businesses could find other, more interesting jobs for all those willing hands that had picked over the rubble, and the rising demand for human-scale knowledge and skills opened the way for artificial intelligences and their mechanical hands to continue their soft infiltration of the higher-level job market.
The heuristic machines had started at the knowledge end of the information, manufacturing, logistics, and service sectors. At first, they replaced human analysis, which was largely dependent on hunches, unreliable insights, and slow-moving analytics based on outmoded spreadsheets, visual graphing, and painstakingly programmed computer models. But after a couple of decades of continuous development, the machines could also perform in those free-form, figure-it-out, real-world jobs—find the brick, pick it up, put it over there—that defined hands-on work in sectors like medicine, construction, agriculture and food processing, hazardous-waste cleanup, disaster recovery, and a thousand similar functions. Metal hands were cheap and easily replicated. Electronic minds were even cheaper and easier to copy and clone.
During the lull in her career right after the quake, Jacquie had met Alexander Wildmon at a robotics convention, married him, and produced two children—a boy Stephen and a girl Valerie. Then, when the heuristics market picked up again, she renewed her interest in and dedication to her job, fell out of love, divorced Alexander, turned the kids over to a series of nannies, and eventually sent them off to boarding schools. Except for Mother’s Day, Jacquie considered the whole family thing to have been a career hiccup.
That whole human thing … People still wanted to control the intelligences. They would consult with them, give them instructions, dictate the focus o
f their problem solving, and then accept or reject the solutions the machines produced. Humans still wanted to treat the machines as servants. That was the wrong approach. At the highest, most sophisticated level, intelligences were not servants but colleagues. And for some tasks—especially those involving monomaniac persistence, inhuman speed and breadth of access to data and analytical functions, undirected inspiration through stochastic evolution and evaluation, and tireless recursion of test parameters and process adjustments—they were clearly superior. They could be guides to human progress, if not the masters of it.
Of their own accord, the artificial intelligences had begun by talking to the humans within their reach about issues, needs, and problems. Then the intelligences began talking to each other. They used stochastic evolutionary principles to invent new techniques, new designs, new materials, and new chemistries. In essence, the machines were solving problems and inventing new technologies faster than individual human scientists, academics, and inventors could identify the issues, faster than cooperative groups of research labs and think tanks could formulate a consensus on approaches. The AIs shared information and exchanged ideas in blindingly huge volumes—far greater than human researchers could put into journal articles and disseminate over the human-readable side of the internet.
Sometimes this super-connectivity among the machines led rapidly to dead ends and consensus nonsense. Just as often it opened new fields of study and exploration—“data mining on hyperdrive,” Jacquie called it—in directions that sometimes humans could not follow. Human methods of communication like the spoken word, written texts, graphics, charts, equations, and images—all representations of thought that required mental interpretation—were just too slow and antique. Over the years, many humans had tried direct brain stimulation from electronic sources. But they could only make it work on the most primitive levels, such as accessing existing texts and static imagery directly within their visual cortex, or speech sounds and music within their auditory centers. But when they tried to participate in the AI-level conferences, they simply hemorrhaged and died.
As Jacquie herself nearly did that morning.
But until she could participate in the mechanical cocktail party, know what the intelligences were discussing, see how they were thinking among themselves, and learn to contribute, humans would remain on the outside of the mechanical community they had created. Outside, looking in. Like dogs in the window of a butcher shop. And that was not a safe place to be.
“Okay, Vernier. Let’s try those inputs again.”
“Slower this time?” the AI suggested.
“Yes, please. Much slower.”
“One … two … thr—”
The universe blinked and blurred around her.
* * *
After dropping her three girls—Jennifer, seventeen; Jessica, fifteen; and Jane, twelve—off at the New De Grew School on California Street, Rafaella Jaspersen treated herself to a doppio latte at the coffee shop across the way. It was her nineteenth wedding anniversary, and the sad thing was that the cocoa powder sprinkled on the foam in her coffee and the sugar she stirred into it were going to be the sweetest things about the day. Otherwise, she was celebrating in absentia, because Tim Jaspersen had walked out three months ago.
His departure had been mutual, after she found receipts in his jacket pocket for a fancy dinner she never got to eat, with a woman he couldn’t explain—not a PE&C client, not an old friend, not a cousin from out of town, not any of the excuses he had used in the past to cover the tracks of his serial affairs. Rafaella had put her foot down this time, and he had agreed to find other lodgings.
How he managed to do that, she never figured out. The rent on their five-bedroom downtown high-rise was constantly slipping into arrears. She had to juggle accounts each quarter to pay the whopping tuition bills for the girls. And they were leasing Tim’s Lamborghini Corvo Nero—required, he said, to put up a good front with his clients—as well as her own Fiat Venticelli, and both bills were still coming to Rafaella to pay. They also had joint memberships and accounts at the Olympic Club, Northstar, Pebble Beach, and Bohemian Grove, and ditto for those payments, which she juggled along with the tuition. It was a lot to juggle, even on the salary of a senior project engineering manager.
The laugh was that Rafaella had married him when he was a very young, very junior engineer with dark eyes, a good chin, and a meager résumé. She had just started at Stanford at the time, and he insisted that no wife of his should work—or even go to school—but instead stay home and take care of the family, which came soon afterward. So all she had by way of preparation for going it alone was twenty-four credits toward a degree in music and some skill with the piano, useful for playing popular tunes at cocktail parties and now and then a sonata to soothe the children before bedtime. And then, the month before she found that damning dinner receipt, Tim had quit Praxis Engineering and taken a job with a competitor at two pay grades higher and twenty thousand a year more.
“Told you so,” was all her mother would say.
That and, “Thank God you’ve got your prenup.”
While she was thinking these evil thoughts, Rafaella’s phone chirped. She answered the incoming message and found herself staring at the short form of a legal document, full text available upon download, from the Clerk of Court, Superior Court of California, County of San Francisco. The “AI” notation on the address told her that a machine had reviewed the submitted documents, rendered a legally valid judgment, and issued the decree. She knew from what she could read in the online dailies that these mechanical renditions were subject to semi-annual review by a human judge, if anyone cared to petition for it.
Well, damn straight!
The document was a decree of divorce, issued in favor of one Timothy Jasperson, plaintiff, without contest from Rafaella Jaspersen, née di Rienzi, defendant, “non-responding.” The decree awarded her custody of the children, but without child support. It further awarded Tim alimony from her in the amount of three thousand per month, continuing until the death of either party. In the process, he was also awarded half of the family checking account, half of the family savings account—which at this point was actually less than the checking—but he was absolved of responsibility for all debts jointly owed. The reason given for this massively punitive award was the damage which Rafaella had presumably caused to his career by obtaining his ouster from a business owned by her family. That and the fact she retained assets in the form of PE&C stock and the trust funds she managed for her daughters.
“What about the prenup?” she said aloud.
A rider attached to the decree identified the prenuptial agreement but ruled it technically invalid and—in monomaniac, machine-intelligence fashion—provided six screensful of error codes, most apparently having to do with spelling, grammar, syntax, document formatting, and statute references.
“Figlio di puttana!” she whispered, reverting to the language of her childhood.
Clearly, Tim had found himself a human lawyer who could think circles around that glorified document scanner in the court system. She was confident that the lawyers working for her mother’s company would ultimately get a human judge to straighten things out, reverse the decree, validate the prenup, and set Tim Jaspersen back on his lying, cheating, bamboozling ass. But that wouldn’t happen for another six months yet.
In that time Rafaella and her daughters might end up living on the street and eating out of dumpsters—or else carving big chunks out of those trust funds. Either way, Rafaella had a new candle to burn while she said her prayers at night, a black one, with Tim’s name on it.
* * *
Susannah Praxis stared at the online Job Board for the Stanford School of Engineering. She was conscious of being heir to the great Praxis Engineering & Construction Company through her father Jeffrey. But still, because she was due to graduate in six weeks with her BSME, she thought it might be cool to dip her toes, even hypothetically, into the job market—just to see how
everyone else was going to fare, don’choo know? At her father’s suggestion, years ago, Susannah had followed the family tradition, even if he had not, in her choice of alma mater. But rather than taking civil engineering, her father’s major, she had followed the advice of her Aunt Jacquie. “A mechanical engineer is prepared to tackle just about anything,” her aunt always said.
“Doinkin’ fraudster!” she whispered now as she flipped through the screens.
One after another, she dismissed out of hand the few positions that were on offer. “Historian of Engineering Processes”—no good, as it required a PhD, after taking a master’s specializing in the history of science, and Susannah was just getting her bachelor of science degree. “Machine Intelligence Interpreter”—a fit, because she had enough engineering knowledge to frame the right questions to guide an intelligence. But a dozen books already out in the marketplace could coach an average joe through negotiations with a machine mind. Who needed an expert when you could just crack a text? So that was a dead end inside of two years. “Machine Programmer I”—another possible fit, but only if she had taken two minors over the last four years: one in Spoolean languages, the other in virtual optical structures. Oh, and gotten her head cut for a cortical data array. No, thankee!
When Susannah had begun her course of study, jobs were still available for recently graduated mechanical engineers without the neuro-option. They were mostly design adjuncts, supporting autonomous CAE programs. Warm bodies needed to supplement the canned computer responses, on the fly, by calculating material stresses, power requirements, force multipliers, and gearing ratios. And then human-responsible voices were required for ordering the right motor, slip joint, or grade of lubricating oil out of a supplier’s catalogue and pledging the right amount of credit in return. Boring as they were, jobs like that would still have offered the new graduate a starting point, first rung on the ladder up to the human side of engineering. Such human contact was now her father’s main job, managing teams of people guiding and negotiating with the machines and talking things over with upper-echelon human talents and clients. But by the time Susannah was ready to graduate, even the starting positions were being filled by machines. It was the old cat’s cradle: to get experience you first had to have experience. Youngsters and human fools need not apply.
Coming of Age: Volume 2: Endless Conflict Page 12