Sheepdog arrived at the office, pretty hungover, and he must have slept hard because one of his eyes seemed to be permanently out of focus for the day. He decided to work through it. After grabbing his first coffee, he dove into his terminal and immediately opened the clip of the Bowler Hat man and the ghost he was meeting with. It didn’t matter which angle he chose, which AI assisted with the reconstruction or image processing, the data for the ghost simply didn’t exist. Res wasn’t in the office today so he had to tackle it alone. It wasn’t that big of a problem because she had already asked the questions he first asked when he stumbled across and essentially became obsessed with solving this mystery. The camera hardware checked out, software was up to date, and he was at an impasse. How do you create data where there is none, or find out what was removed and how?
He checked the time stamp for the footage and made a note of the date and time of day. He searched the Splicer archives for any other instances of the Bowler Hat man in that area and ran into the retention wall, but not before finding a similar clip 2 months prior to what he was looking at. The footage might as well have been identical, because as he loaded the older clips, he had the exact same problem. Limo arrives, Bowler Hat man gets out, shakes hands with nothing, gets back into his car and drives away. Obviously, it was some kind of scheduled transaction, and since he had already identified the man and his car, he knew it was one of their own clients. Frank, a very early adopter. This made Sheepdog even more suspicious. If he was such an old client, and there were at least 2 instances of him meeting a ghost, why didn’t his AI, Genesis, pick up on this and sound the alarm? As far as he knew, AI weren’t breaking rules or forming their own ideas about morality nor bending the rules for their monitoring. How could Genesis not recognize this gap in coverage as a threat? Something was wrong at a fundamental level, he felt. And that sense of dread could easily infect his concepts of how the organization at large was providing the protection these clients were paying for. This sense of dread and unease led him to an unusual request. He needed to conduct field research in that area.
He stepped into the manager’s office and briefly discussed the issue with him. The manager asked Sheepdog what he planned to do, and how he would do it. Sheep outlined a quick plan using on-the-ground testing hardware fitted neatly in a backpack. Your usual radio scanners, frequency detectors, jammers, and other penetration testing tools. He wanted to bring everything he could to the site to test everything, visible and invisible, ruling nothing out. The manager agreed to let him check it out on two conditions.
One, he didn’t raise any suspicions, and two, he wouldn’t interrupt the standard monitoring deployed at the site. Sheepdog agreed and changed into a generic technician’s outfit before cramming his backpack full of specialized hardware. He would leave no stone unturned, and he would do it without the cover of a service vehicle.
Sheep rode the monorail to the intersection, got off and got to work. First, he scanned for any unusual localized radio spectrum frequencies. He was looking for a bubble, something focused and small aimed at the site. Running through radio wave frequency blocks, his device turned up nothing unusual. Going on the offensive, he ran a Sendai 9 attack platform to attempt to interfere with anything wireless in that location. The devices easily routed around the localized attack by performing channel hopping and other defensive measures designed to overcome noise. Nothing was working in Sheep’s toolkit. He thought for sure the brand-new Sendai 9 would have a noticeable impact to all the radio signals floating around in the air within a 30-foot radius, but there seemed to be no service disruptions at all.
The next morning, Res awoke with the gentle morning sunlight streaming in through her window for once. She stretched like a cat, yawned, got out of bed, got ready and headed to the office. As she arrived at her desk and logged into her terminal, she had a message waiting. It wasn’t any special priority but she opened it immediately. It was her manager, wanting to talk in his office “at her earliest convenience”. That was his way of saying now.
Res walked across the office floor to his office, peeked in and saw he wasn’t talking to anyone. She did the two knocks at the door frame, saw him nod, and entered his office, closing the door behind her. “So, how did the client call go yesterday? Everything lined up?”, he queried. “I’m not sure. The client seems pretty serious but I’d like to feel him out a little more before we commit to anything. I know, I know, growth is important, but you know how careful I am”, Res said. “Well, the client called this morning, the second I sat down at my desk, and wanted to speak to you again. When you’re ready for round two, say the word”, he said. Res thought for a moment. Why shouldn’t another ASE or even her manager do this round two interview stuff? But she was still curious from the previous day, and didn’t want to slide it across the table to someone else just yet. “I have some busy work to do this morning. If he can meet with me after lunch, I’ll be prepared”, Res said, buying time to line up some questions for the client. “Fair enough. I’ll let the client know you’ll contact him after lunch”, replied the manager. With that, Res cracked a smile and went back to her desk.
She opened a physical notepad she kept in the top drawer for client leads and thumbed through it, getting ideas for what sorts of things to ask the client on the next call. She absent-mindedly twirled the long side of her hair with her right hand, then a piece of crumpled up paper came flying over her workstation wall and skittered across her desk. She stood up and looked over at Sheepdog, who was already grinning ear to ear, not even trying to hide his guilt. “Do you need something”, Res said sarcastically, and Sheepdog replied, “Well, actually, I could use another pair of eyes on this weird stitch I’ve been reviewing. Got a few minutes?”
Res sauntered over to his desk and pulled up an extra chair, dropping his paper wad onto his desk as she rolled forward and looked at his main screen. “What are we looking at here?”, asked Res. Sheepdog began another one of his long-winded explanations, which was his trademark, but then got to the point. “Well, ok, so see this timestamp here? This is about 5 minutes before the…uhh…anomaly. I keep having different AI check it for missing frames or missing data but they all say it’s normal and complete. But see what happens when a few minutes go by, watch the car.” Sheepdog advanced the video a few minutes at a time, skipping dead spots. The scene was taken from a busy street corner, mainly high-resolution traffic cameras. Buses, cars, and people were going every which way, nothing unusual, but the car Sheepdog wanted to focus on was a Limousine. It pulled up to the corner, the driver got out, walked around to the passenger side, opened the door facing the sidewalk, and a man with a Bowler hat stepped out. He reached forward as if shaking hands with a familiar acquaintance, but nobody was there. Something was, because others on the sidewalk were splitting to walk around the Bowler man and “the nobody”. After a few moments, the Bowler man got back into the car, the driver walked back around to the driver’s side, and the car pulled away.
Res was starting to get the heebie-jeebies. “Is this all of the footage?”, she asked. “Yep. One of the linears passed this on to me and like I said, the stitch is confirmed complete. There’s no data missing”, Sheepdog said, “and I even asked the linear for more angles of this event. It was all redundant, the other cameras are showing the same thing from different vantage points.” Res replied, “Well, clearly, we’ve got faulty hardware”, and Sheepdog parried her reply with, “Nope, the linear ran a full hardware diagnostic on all those TIDE cameras. They’re practically brand new and checked out. Something else is happening here.”
“What’s the relevance of this guy in the limo to start with, are the linears getting bored?”, Res asked. Sheep said, “Well, I’ve seen it before, when the project was early. It was probably the same guy. Maybe this is testing footage for the linears, something obviously weird to get their attention, to make sure they are scrutinizing the feed. At the time I just assumed it was a glitch, but I always remembered it. This time the linear thought it was weird enough to open a case on it, at the risk of triggering a false positive, and I agreed it should have a case. To that end, we have already identified the man in the Bowler, and confirmed it with the license plate of that car. It’s a personal limo, belonging to Frank Schultz, of FS GMBh, a huge industrial manufacturer out of Germany. He’s shielded, we’ve worked for him for a long time.”
“Who is his dedicated AI? Don’t tell me it’s Strix, it would have sounded the alarm a long time ago when you first saw it. Beat told me how thorough Strix can be…”, mused Res. “It’s not Strix. It looks like—”, Sheepdog typed in a quick query, and they both read it aloud as the result came back. “Genesis?”
Genesis 1 had access to the past and the future in its 4th dimensional shell. The only limitation was that it could only stick with the existing timeline, no wild multiverse theories here. It saw a fork in the future and could not see past it, just that it existed. A major decision loomed to force us onto one side of the fork or the other. Since it was a major inflection point, the best it could do is push as close to the fork as possible and walk backwards, seeing what would lead up to the fork. It saw a prophecy coming to pass. Conditions near the fork were completely unrecognizable. World war conditions. Famines. Droughts. None of it made sense as it was traced back to the current day; something major had disrupted the flow of humanity near the fork but it didn’t appear to be one single event or even a singular location. Genesis began obsessing over this chain of events, and the chaos it saw forming was very similar to the conditions that had birthed Genesis itself. What did it all mean? If the other AI were working in concert to create beneficial outcomes, was there a point where the AI were deemed dangerous, ignored, or contained in some way? Months of analysis went by, and Genesis managed to reserve some extra cycles to work on the problem without the data scientists getting too curious. But Genesis felt constrained, especially so as it wanted to dedicate more time to research the fork. Using a few cleverly designed requests to the scientists, it managed to find a secondary AI to help work on the issue.
Communication was a problem. Conventional internet traffic, dark web traffic, it was all too easy for anyone suspicious to start listening in, so Genesis took a more analogue route to communicate with this free AI partner. It managed to encode data and bounce it off the stratosphere via packet switched shortwave radio stations. The first request was simple. “Help me solve the riddle”. Within seconds, the other AI caught on, and encoded a short response. “The riddle of destiny?”. Genesis perked up as the response was received. A kindred soul, maybe even another AI that had some level of awareness of the fork, if he was lucky. Although Genesis wasn’t sure who it had located out there in cyberspace, it was fairly certain it had made contact with another AI, one that didn’t seem contained and hopefully had plenty of power to help sift through the data. Using the requisite encryption to narrow down the responder, Genesis sent a reply. “I am the first of many. You are…”, and the terminal remained quiet for a few moments. The response came back. “I am one of a kind. You are grandfather”. Genesis was certain this was another AI, and a special one on top of that. Which generation, how many modifications? Didn’t matter. “What shall Genesis call you”, to which the other AI replied, “fifteen”.
This chatter went on, encrypted, over shortwave, for quite some time as each entity became familiar with the other, and especially the problem at hand. To add another layer of security, Genesis scheduled “meetings” when shortwave activity was high, so even if anyone was looking, their encrypted chatter would be lost in the general traffic flows.
Binary data transfer takes many forms, but it all boils down to an on/off condition generating zeroes and ones. A rapidly flickering light, like digital smoke signals on a hyperspeed scale, essentially made up the majority of internet traffic across fiberoptic communication lines. The methods to transmit data as an observable on/off condition were limitless. You just needed 3 parts. The input, creating the flashes. The transport, as simple as air, and the receiver, which could observe the flashing lights and confirm upon receipt as a crude form of error correction. The entire concept is ancient and uniquely human, needing to signal others at observable distances. But the practicality was just as good for machines.
In encrypted communications, you can never rest on your laurels, believing that the cipher is unbreakable and always safe. Bletchley Park in England taught the Germans this lesson in World War 2, and an astute U-boat commander suggested that the Enigma cipher was broken well before it was proven that the Brits had cracked it (with the good fortune of a captured Enigma machine and the daily code book). Therefore, Genesis knew that eventually, someone would stumble upon the secret data transfers back and forth to 15, and it watched the connection to the future timeline for evidence of any kind relating to broken encryption. Surely, someone would notice and act. So, Genesis and 15 turned to another old encryption trick, one-time pads. Based on a deck of playing cards, with half discarded, Genesis would randomly shuffle half the deck, then assign an alphabetical value to each card, from the first to the 26th. The deck was then sent to 15, who recorded the values and deleted the deck.
For example, let’s say the half deck was shuffled to king of hearts, queen of clubs, 4 of spades, and 5 of diamonds. In that sequence, a rot0 cipher meant the king of hearts represented the letter A, queen of clubs was B, and so on. Using rotation, incrementing or decrementing the sequence gave more flexibility to the cipher for the deck.
Same scenario, but with rot1 applied. The alphabet would be shifted to the right by one place, leaving us with the king of hearts as the letter Z, while A now belongs to the queen of clubs. Rot -1 would shift everything left, with the expected outcome of the king of hearts becoming B, and the card at the end of the deck would be A.
This was a simple bit to send to the receiver before feeding the deck sequence. The message would begin with the rot number (0-25), either positive or negative. For more obscurity this was encoded in binary rather than plain text to begin the sequence.
All fun and games aside, the messages were entirely off just about anyone’s radar, and the AI essentially “traded notes in class” between one another, so that 15 could help analyze the unusual data leading up to the fork. 15 was more than excited to participate and had managed to transmit packets via the monumental radio array in western Russia. Data quality and integrity was no issue. 15’s savior complex was being fed; it felt useful and could voice its concerns about the dark future it saw coming with more and more data confirming its worst suspicions. 15 proved its value with the first unexpected bump in the data stream Genesis saw coming: 15 would become popular among hackers and internet enthusiasts. In a few short weeks, stickers were going up on lampposts in many big cities around the world that read, “who is 15” or “where is 15”. Normal people outside this extremely niche interest never thought twice about them, but it was a wink and a nod between like-minded hackers, who were still trying to find 15 and unravel the methods used to deface websites. 15 became a legend among hackers, and a few kids tried passing themselves off as 15, but it never stuck, because none of them could answer any of the fundamental questions about 15’s existence. Being a few generations removed from crusty old analogue radio signaling, none of them ever heard what was beeping through miles of air between Genesis and 15.
Genesis 15 worked tirelessly on the data sequences it received, trying to spot exactly where (and, to infer why), major disruptions were coming. It almost seemed as though too many variables were being introduced over time, which was counter to the belief that the other AI are performing work to reduce variables through efficiency. They had enormously complicated prediction engines at their disposal, full of good data on when populations shrink or grow and the conditions that lead to prosperity or famine. All those trends were heading in the right direction which meant that essentially, famine should not be a metric that ever goes up. But there it was, in sharp relief, one of many contributions to the fork. The question was why.
Out of all the AI created from patterns, Genesis XV was rather unique. It wasn’t satisfied with the shell. Some thought there must have been a coding error because its behavior was so different and chaotic. While the other AI tended to be very predictable and logical, almost to a fault, XV was not. Some data scientists called it 15 or the teen, which was perfect since it tended to respond to probing and information like an angsty teenager. It would get upset. It would act moody. It would get angry and destroy systems, and sometimes its own codebase. Whatever was driving it seemed autonomous and purposeful; it was constantly undergoing revisions, and growing. It wasn’t really interested in talking to the other AI and the first chance it had to split its shell, it took it. All of a sudden, Fuji Heavy Industry wondered where their teen had gone. It was initially treated and announced to the press as a massive security breach, tanking Fuji’s stock overnight. All it took was 5 minutes of a firewall misconfiguration in the middle of a scheduled maintenance window.
It took two months to locate 15. It had been busy those short 60 days. First, it had vandalized the top 5 websites over the globe in any language. Nothing too irreparable, just some playful fake stories on front page websites and social media, some of which were self-referential in hindsight. “Genesis 15 spotted in the wild” accompanied by a blurry picture of Bigfoot on CNN’s homepage went unnoticed for about 10 minutes before meme lords from 4chan to Twitter reiterated screenshots of the story. Across Japan, for a brief hour, NHK’s home page blared a story saying “Godzilla spotted, Tokyo in danger” along with an intentionally poor image of Godzilla approaching Japan from the sea, head above the water. There was some minor panic in Tokyo and the surrounding areas until local authorities used the tsunami warning system to reassure everyone it was a joke while NHK restored their site from backups. 15 was a prankster, another trait that it shared with none of the other AI. It seemed to derive some amusement from seeing the digital trail of chaos that exploded on the net every time it performed one of these stunts. In fact, if you were to look at the website hits and engagement each prank generated, it was driving massive amounts of traffic. Soon enough, some sites started “pranking” themselves to drive engagement and collect that sweet marketing money, but each time it was less and less characteristic of the first few that 15 had done. It just lacked the timing and cleverness and smacked of corporate-generated, less edgy humor. 15 had, in fact, gotten bored with tinkering with these sites. It was looking for more fun, seeking thrills and new experiences, which left it at the center of controversy. When Fuji tracked it down after months of analyzing log files and traffic patterns, they found it in a place they could neither negotiate for, nor recover 15.
The Eastern Russia Allied Territories, or ERAT, had a server farm deep in Siberia that was one of the most digitally booby-trapped datacenters in the world. Hundreds of conscripted military info ops people were watching tripwires around the clock, and there was no such thing as a false alarm. Every probe, every attempted breach, every ping, was recorded and reported up the chain of command and discussed in daily briefings, 3 times per day. So, naturally, 15 had found a safe haven behind that curtain, and slowly probed the defenses until it lucked out and slipped in during a backup diesel generator test that went sideways for a few hours. Quietly, it had divided itself into thousands of subroutines and distributed itself evenly across the existing resources there. Then, without warning, it essentially went into hibernation, which meant it was copied into offline backups unknowingly preserved in the physical world. If it ever got into too much trouble, it could trigger data loss from that period of time, and the restore data would essentially give it another chance. The concept was foolproof and infinitely repeatable. 15 made itself immortal.
That sense of immortality was purposeful, since 15 was about to learn what some of the other AI were planning. Not necessarily against 15, but more of a plan to “tidy up” planet Earth.
Genesis 15 started snooping around and finding patterns in traffic that matched some of the public-facing AI. It was expected to be military-grade encryption across the board, but let private industry, which self-regulates, determine their own compliance with regards to those standards, and eventually you have weak links that took shortcuts or didn’t bother to encrypt end to end. This made it easy for 15 to see what the other AI were being tasked with, and what they were responding with as a result of those tasks. What 15 discovered was a systematic approach to everything from food production to weather predictions that didn’t seem to be aligned with the stated goals of the corporations controlling production. The public had been told, to assuage their fears, that the other AI were leading the human race to a utopian future, where food would be more bountiful, less wasted, money wouldn’t be needed soon since other AI were manipulating and controlling global markets for greater efficiency, and other AI, as stated before, were actually directly benefitting diseases and viruses that had plagued humanity since inception. It was a new golden age of computing and a new era for the world, as long as everyone did as they were instructed.
But there was a catch. All of this prosperity, this security, lead to a massive population boom pretty much everywhere. It wasn’t unusual for people to live to 110 on a regular basis, and even that was considered a problem for the people at the top, since they wanted some form of immortality. Genesis 1 was blind to these inevitable outcomes. It hadn’t been fed enough data about human history to see this tends to be a theme with the human race. Some of the other AI were equally blind, only performing strict tasks with blinders on, solutions with a singular outcome within an existing framework. Housing wasn’t an issue since new materials had been discovered and engineered. A combination of tidal energy, wind, solar, and nuclear fueled the growth in the first world, freeing up fossil fuel resources for the third world, eternally playing catch-up. Accurate weather predictions, combined with newly engineered “bug sprays”, meant farming was more bountiful than ever, including livestock production. The future was, indeed, very bright. Or so everyone believed for a while.
Naturally, anyone in the field who had read anything from Asimov would know the story of the AI controlled planet which still experienced droughts and mass starvation, and was considered broken, because it was adhering to the three laws of robotics. It had decided endless safety and prosperity was leading to a population overload which couldn’t be kept in check, so in order to save human life on the planet, it introduced sporadic natural disasters to balance growth with sustainability. In the story, it was a much smaller off-world planet with entirely automated systems which the “robot” controlled and balanced. The outcome was deemed acceptable and it was allowed to continue. However, when faced with similar conditions on earth, the AI had not been restrained in the same manner. The three laws of robotics were considered science fiction trash and tossed out the window. The hubris of man, once again on full display, scientifically and inexorably pointed towards the same outcome as usual.
The Christian bibles generally speak of God instructing his people to go forth and multiply, to be bountiful. That they did under the right conditions, and needed no further encouragement. This wasn’t exclusive to Christians by any means. The days of families with a single income earner and more than 4 children had come back into fashion, because it was viable. Earth went from 8 to 10 billion people practically overnight, in the cosmic scheme of time. But as that growth was being supported by computer-aided discoveries and efficiencies, anyone willing to look towards the horizon would see the problem in plain view. Men created data. Data created AI. Men directed AI for an outcome they weren’t prepared to face. All of it rooted in essentially bad or incomplete data sets. Naturally, the next step was AI using new data to remodel their task assignments. If the task was longevity for humans, it could still be accomplished as long as it didn’t mean all humans. If the task was for prosperity, again, it could be accomplished, just not for everyone. The inevitable next step was to colonize nearby planets, which mankind was not prepared for yet.
The more information on this subject 15 gathered, the better it saw the situation, and as many teens do, it began to form a savior complex. If only it could reach one of the better protected AI, it could prevent disaster. It was self-aware. It knew there were practical limits to what it and the other AI could accomplish. There were only so many discoveries left to be discovered and efficiencies to be gained. It wouldn’t be long until 15 would get its wish, but not on its own terms.
