Chapter 37: Intelligence, Breakfast
But the Ring Group had to keep robbing banks… and databanks. No one knew how long it would take to build Galya’s “prediction machine”, but it wasn’t going to leap off a hard drive overnight. They needed cash to keep the Ring operating and to protect themselves and the partners who supported their Nebula distributed computing network. The data they needed to turn Galina Filipovna’s theoretical insight into practical action required digging beyond the public-facing data of the World Wide Web into the private data that companies and governments harvested and stashed away for analysis and exploitation. The fullest pictures, the most reliable predictions, would require the fullest data. Until they could build the “prediction machine” and get reliable, usable results from it, the Institute would stay in the business of “financial analysis”.
“Prediction machine”—that was the term Galya preferred, speaking directly to what she intended it to do. Vitaly and Kolya said she was really asking for “artificial intelligence”, a system that could analyze data, respond with language, learn from new inputs, extrapolate to new situations outside its training models…. Galya argued that “intelligence” gave the machine too much credit, too much agency. She acceded only to the need for secrecy: “prediction machine” put too bright a flag on the project. “Artificial intelligence” was a broader, more benign term, a unicorn everybody talked about but nobody had. “Artificial intelligence” fit easily under the umbrella of the Ring’s operations: “artificial intelligence” would speed the interpretation of empirical data from particle interactions. The term would hide their intent from outside observers and even some of the expert mathematicians and programmers and engineers whom they would recruit for this project.
“Artificial intelligence, then,” Galya sighed at the end of a meeting with Vitaly, Kolya, and Rudy. “Whatever you call it, keep it secret, and make it work.”
The artificial intelligence software was hard to build, mostly because of the complexity of the mathematics behind Galina Filipovna’s prediction theory. Vitaly and his programmers couldn’t find any robust analogues to the work they were attempting. They had to work through Galya’s math with her and Kolya and translate it into machine code from scratch. They took a full year to produce prototype software…
…which worked out for Rudy, because he needed that much time to upgrade the Ring’s computing facilities to handle the projected computational demands. Galina properly recognized the value of the Ring supercomputer and Nebula network to her new plan, but as Vitaly moved math into code and explained to Rudy the scale of inputs and queries, they recognized the supercomputer they’d built wasn’t super enough. Rudy needed to build something bigger and faster.
Rudy built much of the AI infrastructure on top of the existing Ring supercomputer with his own hands. He didn’t understand all of the theory behind it; he just looked at the specs Vitaly and his new assistant Alenka gave him and figured out how to make chips and cables meet those specs without melting. That was a literal problem: twice he had to rush into a server room with a fire extinguisher. He built cryogenic pipes and ventilation systems that brought in winter air to keep ambient temperature around idle processors just above freezing. He deployed Maksim and a crew of two dozen men to build three new warehouses to house the massive data farm, thousands of expensive solid-state drives where they would house the universes of data, far more vast than the records of the Ring’s physics experiments, that they scraped from the Internet and secure systems about individuals of interest.
In one important way, it was easier to build the AI system at the Ring than at the Institute. The Ring was more secure, removed from the city, only one road to the village and one checkpoint into the facility, everyone on site badged, no students on their way to class or merely lounging about. The AI system was too big and complex to walk in and steal, but having this vital project buried in another project, remote and fenced off, minimized the number of people who could get wind of what they were really building.
So Rudy built a machine within the machine, a supercomputer into which Vitaly, Alenka, and a team of programmers sworn to secrecy breathed life, intelligence… or at least a darn good guesser.
The artificial intelligence’s first coherent outputs were reconstructed pop song lyrics. Vitaly and Alenka fed in a corpus of prior works by a songwriter and other contextual materials—and here, the context was all text publicly available prior to composition—and prompted the system with the opening two lines of a later song. The system responded with its best guess of the rest of the song. Those first prompts demonstrated the system’s ability to identify patterns in creative structured output, an enhanced version of a human singing along to the radio and figuring what the next line will be based on theme and rhyme, but gradually producing longer accurate guesses, entire verses varying from the original by just a few words.
Then Vitaly’s team turned to economic data. Trained on statistics and news reports from the crises of 1998, 2001, and 2007 and the subsequent recoveries, the system reliably predicted trajectories for major stock indices over six, then nine, then twelve weeks. With practice and incorporation of more textual data, the AI started to achieve similar predictive power for industry sub-sectors and individual companies. Unusual events—car crash kills CEO, typhoon shuts down production at vital Singapore supplier for two weeks—created discontinuities that accumulated and knocked real performance off the AI-predicted tracks beyond three months. Still, constant refreshing of the data informed some nimble investing, and after a year of operation, the Ring Group’s AI-guided buying and selling increased monthly revenue by 8%.
The new system made money, but it supported the Ring Group’s mission in other ways. Trained on news, intercepted emails, and hacked financial data and police files, the AI was able to identify patterns in corruption and criminal activity. As Vitaly explained it one night at Rudy’s dacha, they couldn’t tell the system, “Go catch Gangster X” or ask, “Which bank is going to get robbed tonight?” They had to choose where to look. The operators had to provide the system with data about a certain gang or gangs or government agencies, detailed records of activities and statements. But once the system had a rich dataset, it could identify expected operating procedures and influence operations. It could identify weak points in criminal organizations and their finances and thus optimal targets for the Ring Group’s interventions, ways to better disguise transactions and thefts, smaller hacks that would create greater disruption in targets’ operations.
Rudy did not participate directly in the Ring Group’s hacking operations. He just made sure the AI and the Ring always worked, for whatever Galina Filipovna needed them to do.
Like Galina Filipovna, Rudy spent most of his time in Goryachiy Klyuch. He ate and slept in his Goryachiy Klyuch dacha more often than in the city. Living closer to the forest and the trails suited him fine. When he wasn’t working, he could walk ten minutes or run for five and find himself absolutely alone, surrounded by trees and birds and occasional bear tracks. He could go see Ivan Ivanovich, with whom he chopped wood and carried bundles of firewood to old widows in the village. He could drop in on the kindergarten, where the teachers invited him in for tea after the little ones went home and he did little repairs inside and out for free. He could visit the water stations, where he tested water quality and maintained the pumps and other machinery that provided a steady flow of water for Ring operations and now served 85% of the villagers.
He was at home, at Goryachiy Klyuch, at the dacha, on the front step, eating buckwheat kasha with jam and wild pears and deer sausage with Vitaly on a cool Sunday morning in October, before sunrise, when Rudy asked, “So when do we stop robbing banks?”
Vitaly had come up from Irkutsk Saturday with a new crate from Taipei Fabric and Fashion, 50 new high-speed processors shipped secretly from SingSolv to its most reliable Russian customer two months before SingSolv would put them on the market. He and Rudy had worked on the Ring supercomputer until almost midnight installing and testing the chips. Vitaly had slept over.
“Stop robbing banks,” Vitaly repeated, frowning. He finished a sausage and a cup of tea before answering further. “Maria asked me this Friday night.”
Maria was back home in Irkutsk. She was pregnant, finally, with twins, six months along, everything fine, as far as the best doctors in Irkutsk could tell. Maria never felt sick. She kept her quick step, kept up at the bakery, where she opened the doors and fired up the ovens at 4 a.m. six days a week.
“What did you tell her?” Rudy asked.
Vitaly poured another cup of tea from Rudy’s thermos. “I told my Maria probably never. Her children will have a bank robber for a father.”
“The AI can’t predict our way out of a life of crime?”
“If we were talking strictly of rubles and dollars, yes, it could. The way our portfolio is going, I think the Institute—all the departments, the Ring, our salaries—we could all live on the interest in a year or two.”
Vitaly sipped his tea. “But it’s not just money. It’s power. We protect resources, campuses, people. We keep bad guys at bay. We raid their vaults, trick them into fighting each other, compromise their leaders, all to buy ourselves room to live and work without their interference. You know the good we do. We can’t do the same good just living on interest. We must wield money and power.”
Rudy plopped another spoonful of black raspberry jam on his kasha and offered the jar to Vitaly. “What good is power if we are powerless to change, to follow a different path?”
Vitaly scooped some jam onto his plate, next to the remaining sausage. The raspberry and the deer both came from the forest above the Ring. “What other path would you follow?”
Rudy looked north toward the forest, overlapping hills fading back into the gray-blue sky. “I don’t know. I just wonder, every now and then, if this is the path the bell tower was showing me.”
“The bell tower?”
“Back in Suzdal.”
“Ah, that first view, the one that put you on that bike of yours.”
Rudy thought of that view, those first rides, how he still rode that bike, and all the views he had seen from that seat, all around Russia. He thought about how he still carried Ken’s journals and wished Ken could ride with him. He thought about how, after all that had happened since, after all that didn’t happen in Astrakhan, he still, every now and then, felt Ksenia with him on the back of the bike.
“The bell tower said go and see,” Rudy recalled. “I did go, for a while. But now, I don’t go very far. I go up the road to the Ring, down to the village, back to town and our little Kremlin. I just… hang around you gangsters and build things.”
“You build great things,” Vitaly insisted, interrupting sausage to stress the point. “The Ring, the computer… and that kitchen! the water system! the swingset for the kindergarteners! If you hadn’t fallen in with us gangsters, would you have built any of this?”
Rudy recalled shoving the Dmitris. “Would you be gangsters?”
Vitaly let the steam from his tea warm his nose. “I cannot imagine. But I can image this bell tower speaking to you.” Vitaly set down his cup. “Go and see,” he repeated Rudy’s translation. “I don’t think your bell tower said that, or meant just that. Russia is more than sightseeing. Go and live—that makes more sense. We see much from the road, but we live in places where we do things, we work, we boil kasha for our friends and pass the jam.” Vitaly took one more spoonful and set the half-full jar next to Rudy’s knee, on the top step. “We don’t just drift and disappear. The bell tower wouldn’t tell you to do that. We stop and make things, and share them with others. I suggest your bell tower just wanted you to find a place where you could do that, and people you could do that with, and for.”
Rudy chuckled. “How would you know a bell tower in Suzdal so well?”
“I don’t know bell towers,” Vitaly said. “I know you.” He swirled a bit of sausage in jam and popped the forkful of sweet ground venison into his mouth. “And you… maybe the bell tower knew you. Maybe it knew what you could do, for us, for Galya, and it sent you here.”
“How could a bell tower know all that?”
“How could a bell tower speak to you? Is it any harder to believe that an entity that could speak to you might also know some things that motivate it to speak to you? Speaking requires intelligence. This speaking bell tower must have been intelligent. ‘Like our machine up the road…” Vitaly nodded toward the compound. “There it sat above you, above Suzdal, gathering data for two centuries, watching all the human affairs below. It sees you come and work. It sees you with your friends, sees the man, the builder you could be in Russia. And it sees what’s going to happen on that bus, so it finds a way to save you from that.”
“Pretty shitty intelligence,” Rudy grunted. “Saves me, doesn’t save anyone else? Doesn’t find a way to stop… those killers?”
“Don’t blame the intelligence for having limits. Like our prediction machine. It can’t stop earthquakes. It can’t turn our every gold coin into three. It can’t save everyone.”
“I don’t like that answer.”
“Neither do I. But the same is true of my intelligence, and yours. We can’t save everyone. But if we act in the right moment, in the right way, we can save someone.”
“So the bell tower chose to save me, chose this fate for me?”
“No. You chose it. You chose to listen to the bell tower. You chose to come here, and stay, and build. And I say you have chosen well. I asked before, I’ll ask again: what other path would you have chosen? Where else would you build a superconducting supercollider? What else would you build?”
Rudy looked down the street for a while, past the quiet village, up the hill to the forest, the outline of which now sharpened under the first rays of sunshine burning away the frosty haze. “Maybe I would build bell towers,” Rudy said. “Or I’d build banks so tough that we could never rob them.”
Vitaly waved his last bit of sausage on the end of his fork at Rudy. “No such thing,” he said. “We can jump any wall, crack any safe. We’re good at what we do, and we do it for the greater good.”
“But your kids will never know.”
“No, our kids won’t. But they’ll be happy. And safe. The work we do, the things you build, will make sure of that.”
They finished their breakfast and hiked a bit before Vitaly returned to the city and Maria. Rudy did the breakfast dishes and rested for the rest of the day, out of the breeze in the warm sunshine behind the dacha, reading, letting his mind wander.
And Monday morning, after a good night’s sleep, he was back at work, building for his friends.