“What about the other guy?”
“Professor Harry Seymour was chairman of the school’s experimental aeronautics department. Not as much of his stuff is in application, though there are bits of it in BattleZone and in nonmilitary variations like SafeZone. A lot of his research was folded into all three of the Regis software packages. Like I said, pretty much every automated manned combat, flight, or UAV system we have uses one or all of them. And SafeZone’s showing up in CCTV cameras, new versions of OnStar, autonomous parking programs for passenger cars. Self-drive trains. Man, it’s everywhere. This is the age of autonomy.”
“That’s hardly comforting, Bug.”
“You asked.”
“Okay. Now mix Davidovich into that soup. Could any combination of their knowledge be used to take control of one of any of our drones, or anything with Regis in it and turn it against us?”
“Yes,” Bug said with hesitation.
Church said, “I can see where you’re going with this, Captain, and it certainly gives one pause, but we have no indication at all that this is what we’re seeing. It doesn’t tie into the ballpark.”
“Maybe not directly,” I said, and sighed. “But why would we be talking about this if we weren’t all thinking that Regis in the wrong hands could be very damn scary?”
Neither of them commented.
“What happened at the park? That could be the Seven Kings testing out some new toys.”
Church sighed. “Fair enough. Bug, tell him the rest.”
“What ‘rest’?” I asked.
Bug gave me a truly disturbing little smile. “The really scary part.”
Chapter Fifty
Citizens Bank Park, Philadelphia
March 30, 4:48 P.M.
The two forensics collections technicians who followed Jerry Spencer around the ballpark were both professionals, both top of their game. The woman, Gina Robles, had spent the last sixteen years working with the NYPD and was heading up her own division when she was offered a better job with the DMS. Her partner, Laurence Hong, had been with the FBI for eleven years before getting the call. Neither of them held expectations of being lackeys for someone else.
Both had become just that.
It wasn’t the official designation, of course. Both of them had impressive titles, breathtaking salaries, nice offices, killer benefits packages. The works. Each of them even had their own teams, ranging from secretaries to dedicated lab technicians to field techs. Each of them believed—truly believed—that they could run the DMS forensics shop.
Just not as well as Spencer.
It’s never a fun thing when an expert meets a genius. Robles and Hong talked about it over cocktails quite often.
“This must be what it feels like to be Inspector Lestrade,” said Robles one night as she toyed with the olives in her martini. “You know, the cop who’s in all those Sherlock Holmes stories.”
“I know who Lestrade is, Gina,” complained Hong. “He’s a fucking idiot.”
“No, he’s not. That’s the point. He’s a good cop. A solid investigator. But…”
She left the rest hang that night, but it was a conversation they returned to in one form or another a hundred times.
Now, they trailed the genius and kept looking for something useful. Something that would break them out of the lackey role and remind Spencer that they were every bit as valuable as he was.
It was Robles who spotted it.
Down on the field near the pitcher’s mound. Explosions had thrown debris all the way out here from the stands. Broken and partially melted chairs, shattered concrete, torn and bloody clothing, a baseball bat, trash. Ambulance crews were removing tagged bits of red meat so ragged that they would require lab analysis to identify which parts of what kind of body they came from. Male, female, young, old. As Spencer, Hong, and Robles passed by, heading toward a spot where a piece of what could be a control circuit had been spotted, Gina Robles saw something.
It was broken and covered with brick dust, but it was there.
“Wait,” she said, touching Hong’s arm. “What’s that?”
They both stopped, looked down. Their hearts jumped a gear at the same moment. Robles knelt and leaned forward, studying the shape that was almost completely hidden by dust and bits of rubble. Almost.
But not entirely.
“Jesus Christ,” said Hong, who stood behind her. “Holy Jesus fuck.”
Robles immediately turned, cupped her hands around her mouth, and yelled at the top of her voice.
“Jerry!”
Chapter Fifty-one
Thomas Jefferson University Hospital
Philadelphia, Pennsylvania
March 30, 4:51 P.M.
“It gets scarier?” I asked. “Are you going to tell me that the drones at the park were using a proprietary military program?”
“Not exactly,” said Bug. “If that was the case, we could probably backtrack those drones to where the software was stolen from. No, it’s trickier than that. After Regis was developed and sold to the military, DARPA licensed a stripped-down version of it for sale to commercial markets.”
“What? Why?”
“Money. Piles of it. I mean, are you kidding me? Drones are so hot right now. Everyone wants them, and hundreds of companies are building them. It’s a growth market worth billions, and it’s only going to get bigger.”
“And the FAA and FTC have been fighting this every step of the way,” I said.
“Fighting and losing,” said Church. “Though they thought they’d won a major battle when Congress decided that all drones need to have a reset subroutine that can be activated in case of illegal misuse.”
“Right,” said Bug, nodding emphatically.
“I’m not following,” I admitted. “You’re saying DARPA gave them Regis?”
“They gave them a version of it,” said Bug. “A fragment. Actually, it’s a commercial version of BattleZone that’s been retooled for nonmilitary use. It would allow civil authorities to take operational command of a commercial drone under certain specific events. Homeland worked out the details.”
“That’s not necessarily a good sign,” I said.
“Hardly,” said Church. “Bug and his number two, Yoda, were able to crack the security in under ten minutes. They could take over any drone licensed for business or private use.”
“Shit,” I said.
“Which means,” said Bug, “that if we can do it—”
“Yeah, yeah. The Kings and everyone else can do it.”
“Well,” Bug said diffidently, “the talented people could do it.”
“This is nuts. If those drones have Regis, can we track it to point of sale?”
Bug laughed. “Right now, just about everything has Regis. Every jet, every submarine, every tank has the full military package. All commercial drones have the stripped-down SafeZone. And just about every single drone on the shelves at BestBuy, Target, Walmart, Sears, and Brookstone. Hell, Costco has them. Regis is everywhere.”
I said, “Jesus Christ.”
Chapter Fifty-two
Thomas Jefferson University Hospital
Philadelphia, Pennsylvania
March 30, 4:56 P.M.
Church said, “Captain, tell Bug about how the drone evaded the bottle.”
I did.
“Weird,” Bug said, frowning. “That’s too fast for pilot handling.”
“What does that tell us? Is that the commercial version of Regis?”
“Definitely not,” said Bug. “I’m not even sure it’s the full military version.”
“What do you mean?”
“Well … the software used for some drones has a react-respond subroutine, but it’s designed specifically for the bigger UAVs. Raptors, predators, and the retrofitted QF-16’s target drones, and the experimental QF-16X Pterosaur combat drones. Haven’t seen it on anything as small as a pigeon drone.”
“Could it be done? Adapted, I mean? Is that possible?”
“Just about anything’s technically possible, just not probable. But … to clarify,” Bug said, “is there any chance someone on-site was operating it? I mean, right there in the hall with you?”
“No way to know, but I doubt it. There were plenty of Kings goons in the building, but I didn’t see anyone in what you’d call line of sight.”
“Besides,” said Church, “there would have had to be someone operating each of the drones for that scenario to work. We’d have had some eyewitness accounts, and there has been nothing like that. No, I think Captain Ledger’s assumption is correct. This was the drone itself reacting.”
Bug whistled. “That’s awfully fast. That’s like animal-kingdom fast. Wasp-reaction speed, at least. Perception, threat assessment, and action in a microsecond? Damn. If we’re not talking RPA—remote-piloted aircraft—then we are talking some serious software.”
“How serious?” I asked.
He sucked a tooth for a moment. “Not … sure. From what you described, that pigeon would have had to be operating using adaptive-control techniques. We’re talking software that would allow the drone to learn on the, um, ‘fly’ and then strategize based on acquired data and ongoing variables.”
“Can it evade attack?”
“Up to a point. AI software in the Predator drone is adaptive, and this seems to be, too, but it’s not reacting in the same way. Regis is pretty much the cutting edge as we know it.”
“One way or another,” I said, “the drones at the park had to be using some version of BattleZone, right?”
“You’d have to know computers and AI to understand why I don’t think so, Joe.”
“Then give me the short-bus version.”
“Well, you have to start with the nature of UAVs and the software that runs them. There’s a difference between an unmanned aerial vehicle following a preset computer program and something that actually thinks for itself. Most of what is called self-guiding software isn’t really. Mostly it’s programming that allows for a lot of obvious choices. It’s task-driven. Stuff like fly here, drop this, whatever. That’s nothing really new. I have a Rumba in my apartment that follows a set of programs to clean my rugs. And it has sensors that allow it to perform simple react-and-respond functions like not hitting walls and adjusting suction for carpet and hardwood. But that’s not what you described, Joe. You said that it evaded an object thrown at high speed and then seemed to scan the crowd to assess the best possible attack vector. We could be talking AI complete here.”
“What’s that?”
“In the artificial intelligence field there are different classifications for function, for response, for problem solving, and like that. When a computer encounters a problem that it can’t solve—something that requires human intervention or cooperation—we call it AI complete. Or, sometimes, AI hard. This is when simple, specific programming algorithms aren’t going to get the job done. AI-complete problems crop up a lot when vision is required in order to understand a task. Camera lenses, even those that have thermal scans and that operate in a range of visual spectrums, still don’t do what the human eye and its nerves can do. Same goes with what programmers call natural language understanding. You can program a computer to understand anything in the dictionary, but it can’t interpret inflection, sarcasm, or other parts of human speech. Not yet. So, what you described is something a computer probably couldn’t deal with on its own. Selecting you as a threat, evaluating the potential personal harm of what you threw at it and reacting so smoothly, and then planning a counterattack.”
“Even at computer speed? I thought these machines could outthink us. Or close enough.”
“Ha! Computers don’t actually think as fast as humans. Not even close. Look, computers are calculators. That’s what they were designed to do. Every function they perform, from finding a Web site to playing a game is a mathematical process. This plus this equals that. Computers seem smarter because they can do a lot of calculations at high speed. Such high speeds that it looks like it’s doing a lot at once. But the human mind is the ultimate thinking machine. It does trillions of things at once. Everything from the release of hormones to regulating heartbeat to solving a Sudoku puzzle. All of the functions of cells and organs and proteins and all that organic stuff is happening simultaneously. Computers have been built to simulate that by performing calculations at such high rates of speed that it gets to the same result as fast or faster. But we’re not there yet. No computer, not even MindReader, actually works as efficiently as a human brain.”
“Oh,” I said, more than a little confused. “Are you saying it was pure AI or that AI-complete thing?”
“That’s just it,” Bug admitted, “I don’t know. It’s strange. If this is only the drone, then were not really talking about AI. Maybe what we’re talking about is actually AGI. Artificial general intelligence.”
“What’s the difference?”
“Well, for one thing, AGI doesn’t exist yet. It’s a hypothetical kind of artificial intelligence that is supposed to one day perform any task that a human being can perform. A true thinking robot or thinking computer. AGI is also known as full AI, and it’s a computer mind that crosses the line from ultrahigh simulation of the human mind to something that is a machine parallel. Something that can actually think for itself. Something sapient and sentient. Something that’s self-aware.”
“Bug,” I said, “I know that the drone evaded faster than it should have. So it’s either one of these self-aware computer systems or there was someone at the controls who had some spooky-fast reflexes. Not really crazy about either of those scenarios being the case.”
“No argument,” Bug said quickly. “Tough to know which one, though. Right now, the main focus of drone R and D is intelligence, surveillance, and reconnaissance. However, with the complexity of target acquisition in remote areas like, say, the mountains of Afghanistan or the jungles of Mexico and down in South America, some generals at the Department of Defense have been putting pressure on the guys at DARPA to come up with AI programs that will allow an automated system that can select and eliminate its own targets.”
“How?”
“That,” said Church, “is the question at the center of the debate. The science is called neurotechnology. The argument for these kinds of machines is that they could be programmed with a specific set of the rules of war, which would include facial recognition and other identifying software that would allow the UAV to identify targets with a high degree of probability and then selectively remove them. It’s an attempt to realize true AGI and marry sentient computers to independently operating military machines.”
“You don’t sound like a fan,” I said.
“Hardly.”
“The conspiracists out there,” said Bug, “say that because the government has not officially sanctioned that kind of program, some black-bag organization went off the reservation and is funding it under the table.”
“I want to throw the name Seven Kings out here and see if anyone thinks they’re good for it.”
“It would take their kind of money,” said Bug. “We’re talking hundreds of millions of dollars in R and D.”
“It’s in keeping with their level of sophistication, too,” agreed Church.
“Which,” I said, “makes me want to circle back around to the idea that someone has Aaron Davidovich’s research.”
Bug sighed. “That would seriously suck.”
“The other possibility we have to brace for,” said Church, “is that the Seven Kings have Doctor Davidovich himself.”
I nodded. “Which makes me wonder if we shouldn’t ask our pet tarantula about that.”
“Who?” asked Bug.
“He means Toys,” said Church.
“How would he know? Davidovich was taken a couple of years after Toys, um, had his change of heart. Or whatever.”
I made a rude sound that Church chose to ignore.
Church gave me a considering look, however. “It seems like a cold lead, but I’ll
call and ask.”
Chapter Fifty-three
Boyer Hall
University of California, Los Angeles
March 30, 5:07 P.M.
The gathered students and faculty members joined the president in a moment of silence. On the big screen that covered the rear wall, there was a live but silent feed of people standing vigil before a growing mound of flowers, children’s toys, and photographs.
The president raised his head and said, “Thank you.”
The quiet persisted, however. The hall was packed, and this gathering was as somber as the one on the screen. The continual flashes of the press cameras gave the scene a strange strobe quality that seemed to enhance the stillness rather than add an element of movement.
There were synched teleprompters on either side of the podium, but the president didn’t look at them. Instead, he gazed out at the sea of faces.
Finally, he nodded and began to speak. Ignoring the script. Speaking for once truly from the heart.
“I’m standing with you here,” he said slowly, “and together we stand with the families and friends of the people in Philadelphia who have suffered terrible losses. Just as all America has stopped to look east, to the birthplace of our nation and the cradle of our liberty.”
He shook his head.
“America is the most powerful nation on earth. In terms of our economy, our military, our potential. We all know this. The risk of being so strong, however, is that we sometimes fall into a dangerous complacency. We begin to believe our own fiction, our hype that we are not only unbeatable but untouchable.”
He paused and looked at the faces who watched him.
“After Pearl Harbor, when America was delivered a crushing blow in a cowardly sneak attack, we got up from where we’d fallen, we brushed off the dust, wiped away the blood, and stood together to rebuild, improve, arm, and react. We became the world’s first true superpower. And from then until the end of the twentieth century, we were that powerful. No one struck us. No blow landed on us.”