Chapter 3:  Almost (Isn’t Good Enough)

“Look, Dad!” Marla said, shifting her one year old to her other arm and pointing. “There’s a crow looking at us from the roof across the street. Do you think that could be Julius?”

Frank squinted into the setting sun in the direction his daughter was pointing. Certainly, it was a crow. But every crow looks pretty much like any other crow, at least as far as he’d ever been able to tell. And anyway, as soon as the bird realized it was being watched, it flew away.

“I wonder whatever happened to Julius?” Marla asked.

“No clue. For weeks, he came by every day to mooch for strawberries. And then he never did again.” Frank missed the bird, and wondered, too. But not half as much as he wondered what had happened to the thumb drive the crow had made off with during his last visit. The one with the core files for the most artificially intelligent – and eventually diabolic – software program ever created.

Marla guessed his thoughts. “Do you think Turing could ever come back?”

“I don’t know. And I certainly don’t want to find out.”

“But could he?” Marla said, “I mean, did you save enough of the software for Turing to rebuild itself before you erased the rest of it on the government server?”

“Not easily. Remember that I was able to wipe out the backup copy, too, and Turing was only programmed to maintain a single back up copy.”

“Still, do you think it was smart to take the chance?”

“As things turned out, I guess no, since who knows what happened to the thumb drive. But Jerry Steiner had spent his entire career, not to mention millions of dollars of government funds, creating Turing. I know I probably should have destroyed it entirely, but it didn’t seem right at the time. I was thinking a compromise would be to save the really impressive core logic for study purposes. Or maybe someday the remaining parts of the program could be reused in a new, more rigorously controlled program. Anyway, as you’ll recall, those files got saved more by accident than design.”

Later, after Marla had left, Frank returned to his tiny balcony to catch the last colors of an unusually blood-red sunset fade into night. And there the crow was again, across the street, right where it had been before. Except that now it was a jet-black silhouette against the flaming sky.

Could it really be Julius? Maybe he could find out. Frank usually kept strawberries in the refrigerator these days.

But as soon as he set the bowl of berries on the table on the balcony, the bird flew off as if Frank had set a scarecrow out instead.

Unsettled, he sat down. How long had it been since he’d half-destroyed the rogue AI program that had tried to kill him so it could return to taking over the world? Not long enough. Despite what he’d told his daughter Frank was almost sure the program could never reboot itself from what was left.

Almost.

 

Chapter 4

Let’s Get on With it Then

 

The members of the National Security Council were seated around the table by the time Yazzi entered the room. As expected, he also saw several faces he didn’t recognize, presumably CIA staff that would provide the presentations Yazzi had requested.

“Sorry to keep you all waiting,” Yazzi said. “Dick, let’s get started.”

“Thank you, Mr. President,” Gould said. “I’d like to ask Katherine Bliem to set the stage.”

Gould sat down, and a young woman sitting next to him stood up, holding too many pages of notes. How old could she be? Yazzi thought. Thirty-two at most? One of the things that had startled him when he arrived in Washington was the inverted age structure of the District – half the people who worked for the government below the political appointee level looked like kids to him. But why should he be surprised? It was easy to burn out in most agencies – and anyone could earn a heck of a lot more on the outside after they’d paid their dues. The young woman looked nervous, which also was no surprise, considering her audience.

“Please begin,” Yazzi said,” giving her an encouraging nod.

“Thank you, sir,” she said. “With your permission, I’ll set the stage by quickly summarizing the background for the Chinese actions you’re being briefed on today.”

“That’s fine. Please proceed.”

“Very good, sir. As you know, artificial intelligence – AI – is one of the core industries China is targeting under its Made in China 2025 plan – the initiative aimed at allowing China to dominate ten core industries within just a few years. China’s goals in each identified sector are twofold – first, to dramatically reduce its dependence on other countries in the identified industries, and especially its reliance on US products and technologies. And second, to claim major shares of global sales in each of these industries. Specifically, in each category they want to become one of the top two nations worldwide in technical expertise, patent ownership, and global sales.

“In several of these areas, notably information technology and robotics, AI plays a crucial role. The Made in China 2025 document, which dates to 2015, and the New Generation Artificial Intelligence Development Plan, issued two years later, together map out China’s public strategy for becoming a world leader in AI.

Yazzi nodded. This was old and familiar information.

“Next, there’s the Chinese President’s policy of Military-Civilian Fusion. Essentially, it’s an effort to eliminate barriers between the military and civilian companies. The goal is to give the Peoples Liberation Army – the PLA – unrestricted and ongoing access to each cutting-edge technology as a private company develops it – and not only Chinese companies, but also their foreign trading partners wherever possible. Of course, in addition to that, China has for years been the most blatant and successful practitioner of industrial and military espionage.”

Yazzi frowned. No one around the table needed, or wanted, to be reminded that among other cyber coups, the Chinese had hacked the complete design for America’s most advanced fighter – and then promptly built a China version that was practically indistinguishable from the US original. Enough with the background.

“I’ve been told the Chinese commitment to AI development has soared recently,” Yazzi interrupted. “Why now?”

Bliem had noticed the frown and skipped ahead several pages ahead in her notes. “Sir, we see several non-commercial imperatives driving increased Chinese investment in AI. First, we’ve been uncovering indications the Chinese military wants to use AI much more aggressively. The strongest advocates in the PLA see future wars being fought primarily by drones and other robots rather than airmen, sailors and soldiers. They want to be sure China is way ahead of the United States in those capabilities.

“The next driver,” she continued, “is an increase in the already formidable desire of the Chinese government to control every aspect of the lives of its citizens. The unrest in Hong Kong and the government’s obsession with controlling the Uyghur Muslims in the northwest provinces has fanned that fire, and advances in technology – especially in AI – provide the means to satisfy that desire.

“As you know,” Bliem continued, “there are more surveillance cameras in China than anywhere else on earth. Face recognition AI is now good enough to let the government use data from those cameras to track people in real time almost anywhere they go. AI can also be used to analyze information from other sources, like Internet and social media usage, as well as data from the millions of new “Internet of Things” sensors the government deploys every month.

“Not long ago, there was no way to make use of so much data. But China now owns 221 of the 500 most powerful supercomputers in the world – close to twice as many as the US. It can use those machines to tell it what an astonishing number of Chinese people are up to any hour of the day.”

“Understood,” Yazzi said. “But why do you mention the unrest in Hong Kong specifically?” Yazzi asked. “How is AI going to help China there?”

“Not so much there, Mr. president, but within China itself. The government is worried the infection may spread inward from Hong Kong, and they want to get ahead of that possibility. China sent in tanks and soldiers to put down the demonstrations in Tiananmen Square back in 1989, and hundreds, if not thousands, of citizens were killed. Strong as he is, the Chinese president knows the foreign and domestic costs of doing something like that today would be unacceptably high. AT the same time, his inability to stop the demonstrations makes him look weak and powerless, something that must be driving him crazy.

“That’s where the focus on AI comes in,” Bliem continued. It’s as important to the Chinese president for people to know he can monitor their every move as it is for him to actually have that level of information.”

“That plays nicely with his new Social Credit System, doesn’t it?” Yazzi said.

“Exactly so, sir. Every single person in China, and every business, is now rated for good behavior. If someone’s score is high, they get advantages, discounts and other rewards. If their score goes down, just the opposite. And their score can go down for something as minor as jaywalking.”

Yazzi glanced at his watch. “Thanks for the background,” he said before she could continue. “Dick, let’s move on to the next item – how we compare to China in AI research and development. I assume we’re still ahead of China at this point?”

“The answer is yes and no, sir,” Gould said. “We’re way out in front in areas like voice recognition and profiling. The problem is that all the private sector R&D is naturally targeted at achieving commercial advantages, like pushing ads more effectively and allowing people to place voice orders over smart speakers. Some of the underlying advances can be used for military purposes, but it would take a great deal of additional work to do so, even if we had access to all those trade secrets. Which, by the way, we don’t, because they have so much commercial value.

“So, on the military side, we believe we’re in danger of falling seriously behind. True, the defense contractors are happy to sell us whatever we want, but companies like Facebook and Google own most of the patents on the technology we want. Plus, the defense contractors have a hard time competing with Silicon Valley giants for the best AI engineers. And they certainly don’t have the kind of corporate cultures that make rapid development possible.”

“Yes, well that’s our own fault, isn’t it?” Yazzi said. “With all the red tape and complicated contract provisions we insist on.”

“Sad but true, sir,” Gould continued. “So, just to finish up, our usual way to address a situation like this would be to ask DARPA, the Defense Advanced Research Projects Agency, to fund private sector research, and extend grants to universities, to get the work done. But the experts at the universities are aging out, and the big five high tech companies not only hoover up all the new PhD students, they buy up all the promising startups as well as soon as they’re on to something. That’s just as true in robotics.”

“So, you’re saying the companies that have the talent and the cultures to give us what we want have no interest in working for us?”

“More or less, yes, sir,” Gould said. “With the exception of cloud computing, there’s really no overlap. Companies like Facebook and Apple have entirely different business models and have no interest in doing government contract work. And even when it comes to cloud computing, we’ve run into issues where military work is concerned.”

“Right,” Yazzi said, recalling that Google had dropped out of bidding on a ten-billion-dollar government contract after twenty thousand of its employees staged a walkout. “We’re not too popular right now in some tech quarters. But China has huge IT companies, too, and tens of thousands of startup companies besides. Doesn’t it have the same problem competing for good people? Plus, they don’t have nearly the talent pool we do.”

“That’s true, sir. We have roughly three times as many top AI engineers as the Chinese – for now. It’s also correct to say that engineers can make more money in the private sector in China, the same as here. But with a population that’s more than four times bigger than ours, China has the potential to train up a lot more AI engineers than we can. And, finally, China’s military-civil fusion policy means that the military has access to the most valuable trade secrets of private Chinese companies as soon as they develop them. That’s a huge advantage.”

“True,” Yazzi agreed. “So, as I recall, you were going to make a recommendation as well as tell me the bad news. What is it?”

“Basically, it’s to work the other side of the public perception situation to our advantage to get access to the top US AI talent. All of the big five high tech companies – Facebook and Google, and to a lesser extent Amazon, Apple and Microsoft – are getting heat from their customers for not taking better care of their personal data.”

“So?”

“Those same customers are starting to ask their elected representatives to stop the dominant IT companies, and also the big retailers and financial companies, from selling customer data, and do a better job of protecting it, too. A lot of voters think the big tech companies have become too powerful and need to be more closely regulated. Some people even think one or more of the big five should be broken up.”

“Fine,” Yazzi said. “So, I see the possibility for some leverage. How do you suggest we exercise it? I don’t see how we can take advantage of public pressure to make companies do things their employees – who are also voters, by the way – don’t want to do. Or how we’re going to get Congress to stand up to an army of high tech lobbyists.”

“That’s all correct, of course, sir. But despite that, there’s enough bipartisan support for cracking down that the high tech companies are worried.”

“What specifically do you have in mind?” Yazzi asked.

“Our thought is to throw the high tech companies a lifeline while asking for something back their employees won’t object to – something the public should see as beneficial. Specifically, we would identify falling behind in AI as an important national security risk and ask the private sector to step to the plate to help avoid that outcome.

“You could kick this initiative off by scheduling a high tech business summit at the White House, like President Obama did early in his term when he was putting technology policy together. Business leaders love to be invited to the White House under any circumstances and kicking things off that way would send the signal that government and industry are working together to solve a common problem rather than spending years fighting each other in court. That makes business look good, and it makes us look good. During that meeting, you can send the signal that if business cooperates with the administration, you won’t pressure Congress to crank up the regulatory machine – in fact, just the opposite.”

“Fine. And what specifically is it we want the companies to do in return?”

“Agree to buy into the concept of an AI Manhattan Project focused on two things: ensuring America’s AI-based defense capabilities are second to none, and coming up with standards to ensure that artificial intelligence, big data technology and cloud computing are only used for the public good. The latter goal is to give the companies cover with their employees for doing business with us. That way, we get access to all of the best commercial talent for purposes that everyone would applaud, and the big high tech companies get cover to start bidding again on the contracts they’d love to get.”

“I take your point.” Yazzi said, “Can you elaborate on why you want to reference the Manhattan Project?”

“As you’ll recall,” Gould said, “President Roosevelt got the best and the brightest physicists in the western world together, along with industry specialists, to design the first atomic weapons. They were brought together in just a few places, notably Los Alamos, and the synergy was amazing. When the war was over, almost all of those same people went back to what they were doing before.”

“There’s no way companies are going to just hand over their best talent, Dick,” Yazzi said.

“Not full time, of course, sir. But we don’t need them full time. If we could get them together for a couple of weeks at a time a few times a year, we could make amazing progress at the conceptual level. And we’d have more companies bidding to take on the work we’re having a hard time getting anyone to take on right now.”

Yazzi mused on that for a moment. It was an interesting idea, at least in concept. “What do you think,” he asked, turning to Abner Capp, his National Security Advisor “Does that make sense to you?”

“It sounds plausible,” Capp replied. “It checks off all the boxes. Everybody gets what they want – the companies can quit worrying about new regulations, Congress can go back to sleep, we look good to the voters, and the decks are cleared for us to make lot more progress on advances that can be incorporated into AI-enabled weapons. We don’t need the corporate employees to be directly involved in that, since we’ve already got the Pittsburgh Project in place.”

“How about the individuals, though,” Yazzi asked. “What if they don’t buy into it?”

“Our thought on that, sir, is to make it very attractive to them. That’s another reason for analogizing to the Manhattan Project. Being invited to be part of an effort we compare to that project would be quite an ego trip – and ego will get you a long way in the high tech community. I’ll wager there will be a lot more people begging to get in than refusing to join. We can also select venues that anyone would want to spend time at, instead of some windy, isolated mesa in the middle of nowhere, like Los Alamos. Plus, there’s the opportunity to get together with the best of the best and strike intellectual sparks – not to mention show off. Finally, there’s the opportunity to make a little history. Anybody in physics can still reel off a dozen names of folks who were part of the Manhattan Project. Most people never get a chance to be part of an elite group like that.”

Gould waited. Would Yazzi give the green light?

He would. “Okay, I’m persuaded. At least to the point of giving you the go ahead to put together a detailed proposal. When could you have that ready?”

“That’s really outside my area, sir. As you know, I’ve got a fancy title but no authority. I’d need to draft in folks from various other agencies to do this right.”

“Got it.” Yazzi turned to Carson Bekin, his Chief of Staff. “Carson,” help Dick make this happen. Shoot for getting me something to look at within a week.”

*  *  *

Author Notes: In this and the previous post, you’ll have noted that I’m primarily involved with setting up the foundation for the story that follows. In these specific chapters, this presents two challenges. The first is how to put out a lot of information without boring the reader (the most frequent criticism found in reviews of my books is that they “start slow”). I use a variety of approaches to try to keep things moving as best I can notwithstanding the unavoidable need to transfer a lot of information.

One is to break up the flow of information by spreading it across multiple scenes spanning several chapters. A second is to convey it primarily through the use of dialogue. The third is to intersperse these scenes with unrelated, more active scenes. The final one, which I’ll continue to work on in each draft, is to trim out as much as I can to reduce reader fatigue.

As there’s still a lot of information the reader needs to receive, you’ll see me continue to struggle with this challenge throughout the first half of the book. One beneficial aspect of this tension is that it keeps me constantly thinking about ways to bring action into the plot to break things up, and this frequently leads to new twists, subplots and characters.

I mentioned that there were two challenges. The second is that many readers are adamant on the subject of “show, don’t tell.” Simply having a character reflect on the information in her head can be a cause of irritation to such readers, as would simply laying factual background out in a way that has nothing to do with a character at all. Conveying the information in dialogue is better, but then a different problem arises: many readers may know little or nothing about some of the topics they need to understand in order to make sense of the plot as it moves forward. On the other hand, many of the characters – in this case, president Yazzi – would obviously already know a lot about some of the same matters. Why in the world then would someone be telling the president what he already knows?

So, what’s an author to do?

One approach would be to use different characters to exchange the information, and often times I will take that approach. Here, though, tossing in a couple of characters to no other purpose would be artificial and intrusive. Instead, I opted to chip away at the issue by having Yazzi’s first briefer be young, nervous, and over-prepared. That provided a cover for laying out more information than the president actually needed to be told. At the same time, it allowed me to introduce some additional human interaction into the scene. Not brilliant, but hopefully sufficient.

All that being said, chapters 2 and 4 are still slower reading than I’d like them to be, so I’ll continue to look for opportunities in later drafts to make these early chapters a faster and more compelling read. While I find this type of self-editing one of the more tedious parts of the writing process, it’s also one of the most essential steps on the road to the finished product.

Next week: A simulated robot vs. human battlefield exercise. Continue reading here

Download the first book in the Frank Adversego Thriller for free at Amazon and elsewhere