by Andrew Watts
“Ha. Yeah, exactly.” Pace smiled.
Nader shifted in his seat. “Well, not war. Learning. Improvement, optimization.”
Colt said, “And . . . that cranial neural interface . . . are you guys actually able to link up machines with people’s brains?”
Pace said, “One of our competitors actually had a chimpanzee control a computer with its mind. Neural interface control would be huge in the medical field. Disabled people being able to use an appendage they could control without physical input.”
Nader said, “It would be incredibly lucrative.”
Colt was incredulous. “How far along is this? Are your results public?”
The two men looked at each other. “We have a way to go. And nothing is public yet.”
Nader looked at his watch. “How about we take this conversation to lunch? I’m famished.”
The three men headed down the stairs and out the front door, each of the executives shuffling away their own set of engineers and scientists who were eager to speak with them about ongoing work.
A white electric SUV was waiting for them outside. It sped them off into the heart of downtown, dropping them outside an upscale restaurant on California Street. A waiter showed them to a corner table covered by a white cloth and decorated with fresh flowers. The restaurant served mostly Chinese-inspired cuisine, but Colt refrained from telling them he’d had similar food the day before. And this place was much higher-end. Soon Colt was drinking hibiscus iced tea with mint and looking at a menu that didn’t have any prices. If you had to ask . . .
“I’ll order for us if you don’t have any objections. They do it family-style here,” said Nader. He ordered in Mandarin.
“You know Chinese?” Colt asked after the waiter left.
Nader shrugged. “I took it at Stanford, and then did an internship in Shanghai for a few months after graduation. The way China’s tech is booming, it’s been quite helpful.”
Interesting. Colt didn’t remember seeing that detail anywhere in Nader’s file. He would have to ask Weng what she knew about his time in Shanghai.
Soon the waiter brought them a circular wooden box of dumplings and sauce. Colt used chopsticks to hold his while biting into it. An explosion of tasty broth, pork, ginger, and spices filled his mouth.
“Oh my God, this is incredible. Do you guys always eat this well?”
Pace didn’t respond. He was thumb-typing on his phone.
Nader laughed. “We’re very lucky to have some nice restaurants nearby. I must admit that as our company has taken off, my diet has improved. My weight, not so much.”
Colt said, “I did examine your overhead on the plane ride over. Pay and benefits for AI engineers is very competitive.”
Nader said, “It’s a bit embarrassing what some of these AI companies will throw at you nowadays. Disgusting sums. But that’s all going to change.”
Colt looked at him sideways. “Are you saying companies will no longer pay top dollar for AI engineers?”
Pace shook his head, looking up from his phone. “No, he means people won’t be paid so differently. AI will completely upend our economy in the next ten years.”
Colt said, “What do you mean by that?”
Nader sipped his drink. “One of the realities our world will have to confront is that more and more jobs will go by the wayside. And not just blue-collar jobs, either. Bots will replace almost all of us. But here’s the thing: how long does it take the average person to learn a new job? Six months? A year? Five years to really gain mastery? AI will learn much, much faster. Doctors, accountants, lawyers . . . they’ll all be replaced. Or augmented, and we’ll need fewer humans in those roles.”
“So . . . you’re saying we’re all going to be out of work?”
Pace looked up, now interested. “Those who keep working are all going to be working different types of jobs. Ones that demand emotional intelligence or a deep human understanding to do well. But technology keeps evolving at a faster rate. Up until our era, the rate of change was manageable. But with AI . . . now you can write some code that displaces millions of jobs within a few days’ time. As machines begin writing their own code, they will displace jobs even faster . . . hell, they can invent whole new industries within nanoseconds. Human beings can’t keep up. It’s like that I Love Lucy skit with the chocolate factory production line. People weren’t meant to work around the clock like they do today. Getting emails all night long. Eyeballs in their phones. Screens and computers pummeling information in and sucking data and money out. But machines can always go faster. Human beings can’t.”
Nader threw up his hand. “Nice going, Luke, there goes our funding. You’ve converted another one.”
Colt smiled, but he had to admit, he felt a little chilled by Pace’s doomsday view. “So, Luke, I thought you were excited about the future of AI?”
Pace sighed. “Oh, I am. I don’t mean to scare you. For most people, the pressure and the demand of daily work will finally cease. Make no mistake, however, universal basic income will be required. The old days of everyone pulling their weight and getting money in exchange for value creation will be gone. People simply cannot keep up with the pace of change. And in a world where humans are surrounded by immortal, all-knowing, thinking machines, why would one person ever make more money than another?”
Colt offered what he hoped was the right answer. “You are suggesting they wouldn’t?”
Pace pointed his finger at Colt. “Bingo. We are ten to twenty years away from living in a world where wage inequality will start to go away.”
Nader snickered.
Colt said, “You don’t agree?”
Nader glared at Pace. “I had hoped to shelter you from Pace’s doomsday scenario.”
Pace looked like he realized he’d gone too far. “Sorry. Maybe I am a bit pessimistic. I have a lot of work back at the office.”
A few moments later, they left the restaurant, passing through a crowd of tourists. Colt scanned the faces in the crowd, trying to maintain good street awareness even in an impossibly crowded environment.
Colt saw her coming.
Marisha Stepanova approached him from the east, masked by the crowd. She wore a beige waterfall coat over a white tee, skinny jeans torn at the knees, and dark sunglasses, and was carrying a Starbucks cup in her left hand.
With her right hand, she executed the brush pass.
She slipped a thin phone into Colt’s right pants pocket as she walked by. He forced himself not to react.
“After you,” said Nader, holding open the SUV’s door for him.
“Thanks,” Colt replied, and ducked as he entered the back seat. When he was inside the car, he glanced out the back window.
SANDSTONE was gone.
Colt waited until they were moving and both men were distracted, checking their emails on their own devices. Then he removed the phone Marisha had slipped into his pocket and tapped to illuminate the screen.
It was unlocked, but required him to set a password, which he did. On the home screen, he saw two messages awaiting him.
M: Cease all previous communications methods. Check this device daily.
He texted a reply.
C: Okay.
He was about to stuff the phone in his pocket when he saw the three flickering dots that signified the other party was typing a response. It came a moment later.
M: I will know more about the leak soon.
12
“She’ll know more about the leak soon?” Wilcox repeated. It was just after eight p.m. local time. The other team members were gathered around the central table in the counterintelligence unit safehouse.
“That’s what she wrote.” Colt nodded.
Weng tapped her pen on a notepad. “And no other information?”
“Just the part about comms being compromised. She wants me to use the phone she provided for all further communication.”
Weng said, “If she’s going to get the identity of the Pax AI mole, does
that mean it is more likely another Russian intelligence source?”
Colt frowned. “Wouldn’t she already have known about it?”
Weng shrugged. “I don’t know. Could be a parallel operation. GRU not wanting to share with the SVR? I mean, how else would SANDSTONE ascertain the identity of a mole in Pax AI unless she had access to information there? And if Kozlov is now dead . . . doesn’t that imply the Russians have another agent inside?”
Rinaldi said, “A logical deduction.”
Wilcox shook his head. “I don’t know if we can trust her at this point.”
Sims said, “At least she’s back in contact. Have you messaged her yet?”
“Not yet. I wanted to update you guys first and go over next steps. I agree with Wilcox. I could be getting played.”
Rinaldi said, “We can get some of the crypto guys to take a look at the phone. You want to give it to me? Where is it now?”
Wilcox sighed. “We should never allow one of our agents to dictate what covcom we use. It’s very possible that a GRU cyber espionage unit is using that device to monitor everything you do. If it’s within fifty feet of you, they’ll know your location, listen to your conversations.”
Colt said, “I know, Ed. I have it in a storage locker across town. I’ll obviously never take it to Pax AI or to any meetings.”
Weng opened a stick of chewing gum and popped it into her mouth. “We can set up countersurveillance on the drop box. That way you don’t get rolled up during your daily comm checks.”
Rinaldi said, “You’ll keep us updated on everything she sends you?”
“I will.”
“We should go over any responses, too.” Wilcox frowned. “I just don’t like this. It would be better to set up a face-to-face, so you know it’s her.”
Colt said, “I imagine she’s under intense scrutiny after Kozlov’s death. Or she’s been loyal to the SVR this entire time . . .”
Wilcox rubbed his hands together. “Bottom line, we need more confidence in the information she provides. She wants you to use that phone to communicate. Fine. But what else can we do to mitigate the risk?”
Colt said, “I have a technique that I’ve used in the past with this agent, when we were in high-population-density areas. We can set up a meeting but be in different buildings, across the street from each other, windows facing, so we can see that neither party has a gun to their head. I can tell her I want to do that again. We can still use her phone but set up cameras to verify the accuracy of communications.”
Rinaldi said, “I can help with surveillance. What’s the name of the storage facility? I’ll send a team there now to watch it.”
Colt gave him the name and Rinaldi turned to Wilcox, who nodded. “Please set it up.”
Weng slapped Colt on the shoulder. “How did the rest of your first day of school go, honey?”
Colt walked over to the fridge and grabbed a bottled water, then sat back down. He took them through every detail of his first day at Pax AI. The morning briefs, the lunch conversations. After lunch, he’d spent the rest of the day being shown around by several different teams, learning more about their projects and future plans.
“Still no fourth-floor access?” Wilcox asked.
“Kim implied it’s coming,” Colt reassured him.
Rinaldi said, “So you think Pace is discontent?”
Colt said, “If anyone I’ve met there thus far is sympathetic to Trinity, it’s him. He was reciting the same concerns I’ve read about in news articles about Trinity. Pace is worried the race to develop artificial general intelligence is dangerous. ‘Playing with fire,’ he said. He seemed particularly worried about their work at The Facility.”
Weng was still chewing gum. “Makes sense. Pax AI’s Mountain Research Facility is the Area 51 of AI development. A lot of highly classified government contracts are underway there. In all the Trinity conspiracy chatrooms, The Facility is ground zero for their impending AI doomsday scenario.”
Colt said, “What kind of projects are they working on there?”
Sims said, “Both IARPA and DARPA have a few. You can use the computer in our back room to access the Congressional summary on those programs. But it’s pretty sparse on details. Most of the programs are codeword-level programs that even we aren’t cleared for.”
“And Pax AI researchers are?”
“Only a few,” said Wilcox. “And trust me, they are all being closely monitored.”
Colt said, “How is that not a security risk?”
Wilcox clasped his fingers. “It’s a technology arms race. Think of it like the Manhattan Project. The US government needs the best scientists in the world working on its behalf, or we’ll lose. And a lot of the best AI engineering talent comes from foreign nations. The same thing happened during the nuclear arms race. A lot of our scientists were under suspicion for being sympathetic to our enemies. Today we have an additional level of risk due to our partnerships with private companies like Pax AI.”
Colt knew he was right.
Rinaldi said, “I’ll put more surveillance on Pace. We’ll have our cyber team go through everything he’s ever typed. If he’s involved in Trinity, we’ll know about it.”
“What about Nader’s Chinese connections?” asked Colt.
Rinaldi turned around the laptop he had open on the table. “His post-college internship in Shanghai was in his file, but plenty of business executives have spent time in China, especially in Silicon Valley. But check this out . . .”
The group leaned in to see what Rinaldi was pointing to on the screen.
Weng hummed. “Very interesting. Gotta love my boy Liu and his MSS college recruiting program.”
Colt said, “So Liu was recruiting agents at Stanford while Nader was getting his degree there.”
Rinaldi nodded. “Yup. And then he gets an internship there. It’s circumstantial, but it’s something. Keep your eyes open, Colt.”
Colt arrived at the Pax AI office early the next morning. Many of the engineers were already there, working on their computer or conducting tests with robotics in labs. As Colt walked to his office, he again saw meeting rooms packed with diverse groups of scientists, working through complex formulas and graphs they had drawn on white boards.
Colt sat in his black leather swivel chair, the glass walls allowing him to see into several of the adjacent spaces. The company CFO appeared at nine a.m. They met for an hour, going through financial data and projections.
After he left, Colt headed down to the break area to grab a cup of coffee. Gerry Nader was there, speaking with two of his subordinates. Colt overheard him saying something about the control problem. The two employees left, and Colt found himself stirring his coffee, alone with Nader only a few feet away.
“Sorry about Pace at lunch yesterday.”
“What for?”
“Oh, he can be a bit of a downer when it comes to conversations on our future. A brilliant mind, but definitely a pessimist.”
Colt thought that was putting it mildly. “Don’t be silly. I hadn’t given it a spare thought.”
Nader said, “So is there anything you need?”
Colt smiled. “Just hoping for that fourth-floor tour.”
Nader said, “I’m told it’s coming. I’ll see if I can push the powers that be a little harder.”
Someone called to Nader from down the hall.
“Gotta run. I’m already late for my ten a.m.”
Colt headed back to his office and continued going through company documents. As much as he was there for the counterintelligence investigation, Colt still needed to perform his duties for his cover job, providing a thorough financial evaluation on Pax AI for deep-pocketed investors. He spent the rest of the morning going through the company’s quarterly reports. He ate lunch at his desk and was taken through various projects by Pax AI employees in the afternoon: autonomous navigation, conversational interfaces, ambient computing, and productivity gains from big data synthesis.
There was
even one on agriculture, led by Luke Pace.
Pace showed him a video of a farm near Fresno, California. “We have automated a lot of the process. Solar-charged quadcopter drones monitor the fields. We use deep learning programs to evaluate what issues our crops are facing. Whether it be pests, irrigation, fungi . . . whatever . . . the drones provide imagery, our AI analyzes the images, evaluates the problem, and sends out a command to fix it. In most cases, that still requires some human intervention. But our team is working on automated robotic solutions to nearly every issue we face. This can increase crop yields and lower costs dramatically.”
Colt nodded. “Fascinating.”
Pace said, “And you can see how the same AI programs, using deep learning to evaluate and learn from large data sets and automated responses, could be applied to many other areas. Take health care. Automated, online diagnosis and prescriptions. Boom. Patient time is reduced, quality and costs all improve. Take computer programming. Our new language algorithms can write code as good as most mid-level coders. We can now tell the AI what our objective is, start writing the code, and they will finish hours of work in mere seconds. We no longer need human coders to slog away through lines and lines of code. Instead, human coders become more like proofreaders or editors, double-checking for errors and quality issues. And the AI will learn from that work too.”
“Rapid continuous improvement.”
“Exactly.”
Colt decided to dig. “It sounds like these programs can replace just about everything. What’s left for us?” He saw that same flash of fear that had been in Pace’s eyes at lunch yesterday. “I’m sorry. I wasn’t trying to say anything controversial . . .”
“Not at all,” Pace said. “To be honest, you hit the nail on the head. It will be a scary transition for people. Until now, technology advanced at a manageable rate. This evolution will be much bigger, faster, and more impactful. Anyone who tells you they know how we will all come out of it is lying, whether they know they are or not.”
A movement outside the glass office door caught Colt’s eye. A man with a crew cut cracked open the door. “Mr. McShane?”