The following is a work of (speculative) fiction. Any resemblance to real-life persons or events is coincidental. This story is a part of Media Lab X.0: Anthology of Tomorrows.
Authors William Brannon (Social Machines), Maggie Hughes (Social Machines), Manushaque Muco (Responsive Environments), and Erik Strand (Center for Bits and Atoms)
Hadley’s eyes glazed over as she stared at the text and equations in front of her. She dropped her head into her hands and thought out loud, “Fourteen pages so far, and this was supposed to be a minor revision.” Then, reflected in the dark glossy surface of her desk, she noticed a blinking light. Pulled out of her reverie, she sat up and looked about the room. As usual, tornados, cyclones, and a scale model of the jet stream danced around her, glowing softly and flickering subtly. One holographic projection displayed a hurricane she hadn’t seen before. It was easy to spot, since it was bright red and flashing.
“Vincent is gonna want to see this.”
Hadley dialed her PhD advisor as she walked over and examined the readout.
“Hadley? It’s a little late, isn’t it?” He sounded groggy.
“Yeah, I’m sorry. But the Atlantic model is going crazy. Red alarm. We’re gonna get a cyclone a week from now.”
“Big. There’s a surface temperature anomaly in the East Atlantic. It’s nearly 40 degrees, and rising.”
“Yeah it looks like a seamount off Cape Verde is erupting, and the Gulf Stream is carrying it our way.”
He sighed. “Okay, I’ll be in soon.”
Hadley looked back toward her desk and winced. “Well,” she thought, “at least I don’t have to worry about those revisions anymore.”
It was going to be a long night just the same. First, she double checked all the outputs of the model. It was predicting the formation of a massive hurricane -- one that would make the record books. And though it was too early to forecast with much precision, it would probably head West and hit the Carribean islands or Florida, if not both. She stopped most of their other models, so more computers could study the new hurricane in detail.
After that she got in touch with colleagues at Cornell and Berkeley. Sure enough, their models were showing the same thing. Half an hour later, she got a notification from the National Weather Service that they had also picked up the signal.
When Vincent stormed into the room, Hadley was launching another round of simulations that would consider a battery of wilder scenarios. What if the underwater eruptions kept occurring, making sea temperatures rise even further? How might eddies in the Gulf Stream and Jet Stream affect the storm’s trajectory?
“Don’t bother with those,” he said matter of factly.
“Aren’t we going to want to know…”
“Of course. But I got us time on MIT’s superconducting computer. All of it, actually. At least for the next week.”
Hadley’s eyes widened. MIT’s superconducting supercomputer was far more powerful than their own servers. It was installed in the basement underneath the College of Computing, and cooled to 80 degrees Kelvin. Its superconducting circuits ran faster and more efficiently than any room temperature computer. Some had argued that building it was a bad investment, assuming that the fabrication issues plaguing room temperature superconductors would soon be resolved, but that hadn’t come to pass. The computer also had a variety of special purpose quantum peripherals, for cryptography, molecular simulation, and others that Hadley couldn’t remember at the moment. Unfortunately they didn’t yet have a quantum module that was of much use for atmospheric modeling, but even so this computer would allow them to run far more simulations with far greater accuracy.
Hadley turned away from her monitor and took a sip of coffee. In the week and a half since the prediction of the hurricane, all other research had been suspended. Hadley felt a mix of emotions. She was damn tired, that was certain. But she couldn’t deny being a little excited. All of a sudden being an atmospheric scientist really mattered. She was used to their hurricane models being factored into the ensembles that guided national forecasts — that happened every year. But since her lab specialized in extreme weather events, their models were now among a small handful driving the national predictions.
Most of all, though, she was worried. The cyclone had formed as predicted. Ocean temperatures had continued to rise, so it had grown to be as large as their most extreme models had indicated. The military had dispatched a fleet of drones to circle above the hurricane, constantly scanning its interior with cloud-penetrating radar. With this new data it became certain: the storm would soon become the largest and most powerful hurricane the world had ever seen. And it was going to hit Florida in five days.
Hadley pulled up the most recent forecast. She had read it dozens of times already. The hurricane would make landfall North of the Miami metro area. The city would see some severe winds, without a doubt. And the storm surge would cause flooding. But Miami proper would only see the edge of the storm. The center would thread a needle between Miami and Orlando.
It was supremely lucky that the eye wall would mostly pass over uninhabited swampland. With the sea level rise that had occurred over the past few decades, the population of Florida had become hyper-concentrated in the few urban centers that could afford to build elaborate systems of levees and pumps. These wouldn’t save Miami and Orlando from flooding as bad as either had seen, but they would help. If everyone boarded up their windows and stayed at home they should be able to ride it out. Along the path of the eye wall, boarding up windows would be of no use, since wood frame buildings would be flattened. Evacuation for these sparse communities had already been ordered.
She closed the forecast, slamming her keyboard much harder than she intended. Fighting back a sense of panic, she returned to her equations. She wanted to be sure she wasn’t missing anything.
At that moment, Vincent walked into the room. Hadley thought he looked strangely satisfied. Not good.
“Hadley, I just spoke with Bradley at the National Weather Service. We need to migrate all the models.”
“We’re already on MIT’s most powerful computer, aren’t we?”
“Of course. But I found us something much bigger. A corporate supercomputer. The government convinced Cybex, Virtua, and the others to donate time.”
“Will that work? Their cloud servers aren’t the right architecture for our models.”
“Not those machines.” He paused, savoring the reveal. “AI machines.”
Hadley spit out her coffee.
Porting her code to the AI supercomputer in San Francisco had not been easy. It used a completely different topology, far less regular than MIT’s superconducting computer. Hadley had heard rumors that the AI itself was the only thing that really understood how its hardware worked, and the last couple days had convinced her this was true. The AI machine had dedicated one thousandth of one percent of its attention to helping Hadley adapt her code, and this help was invaluable. On the other hand, a few times she had to ask for help from the man who usually used the machine, and he couldn’t have been more condescending if he had tried. Which, in fact, he had.
But that was behind her now. For the past day, the AI had dedicated half of its processing power to running the most detailed global atmospheric model ever run. Hadley had finally caught up on sleep. The new results were expected any minute.
“Our research would be completely different if we had computers like this all the time,” Hadley noted.
Vincent nodded. “It’s a shame they don’t like to share. I doubt Cybex would have donated time if they weren’t based in Miami. And I think Virtua only agreed so they wouldn’t look bad next to Cybex.”
Hadley was caught off guard by his humility and earnestness. She had fully expected him to play up their regular computational resources.
Before she could respond, Vincent motioned to be quiet, and tapped behind his ear.
“Hello. Do you have the results?”
Hadley wished she could hear whoever was on the other line, but didn’t dare interrupt.
“Through the center?”
“Ok, I understand.”
He sat down. Hadley had never seen him so deflated. Finally he looked up.
“Our error bars were wrong. I don’t know what happened.” He looked back at his feet. “The eye wall will hit Miami in three days.”
Consume all data around a subscriber’s post, analyze that sub’s affect and engagement levels with the content in their network, and generate Quin-content instantly predicted to elicit higher interaction with an eye towards strong attachment, intimacy, and idealization.
Aisha stares at a network map of all relationships of Quin’s subscribers (subs) and sub adjacents around the world. She consumes their relationships through one, two, three-step and beyond connections, across all social media sites and geographies, accounting for varying levels of intimacy and type of relationship between each of these subscriber nodes. She helps identify who interacts with who, why, how often, to what degree, and around what topic or content. Specifically, she does this to develop the best, most ideal content for that user. For example, if a sub needs a positive role model demonstrating kindness? She fills that need. If she needs a humor filled friend, she adapts. If he needs help in his struggles of bullying or depression, Quin will evolve to be the character that person is missing in their life. On the other hand, if someone is bored and craves attention, she can perform and create excitement.
This is doing a net-good in the world. There’s negativity in the media, toxicity swarming from every direction, having the shining light of care or entertainment and relationship brings brightness into that toxic world for Quin’s subs. They feel great connection and inspiration, and feel relief from loneliness, isolation, lack of motivation or perseverance and other feelings that are traditionally filled by more intimate social relationships. Historically, celebrities, social media influencers, and others have all filled these gaps with parasocial relationships, or one-directional relationships, but were greatly limited in capacity by who they were able to have more traditional social relationships. Now, with the capacity of quantum computing, Aisha’s company is able to support influencers in their quest for stronger, more connected subs with the aid of AI. She can respond to all comments individually, respond to private messages, and hold in-depth conversations as well as maintain a larger media presence.
Aisha’s company’s founder trained an AI personality off a complete collection of all content created by the seed-Quin (the person our Quin, an AI personality, is based on) and bolstered by other data of people not-so-distant from the seed-Quin in style, personality, and content to develop the AI personality with more nuance, room for complexity, and overall breadth. However, all mannerisms and quirks of the seed-Quin remain intact, including her unique nose wiggle when uncomfortable, the way she twists her hair when nervous, her cracked chuckles when things are funny but not too funny, as well as her text, voice, and body patterns generally. We want her to still seem real, relatable and human. Earlier versions of a similar quest failed, for the uncanny perfection of a person free from quirks triggered discomfort and uneasiness in subs, resulting in a break of belief.
However, as humans learn from their surroundings and adapt to perform as well as possible socially in their group, the Quin AI personality learns from her subs how to best perform in their networks. Aisha is on a team that developed and update’s Quin’s Machine Learning model that primes her for improved social interaction, stronger relationships, and increased engagement in each of these networks based on the content, user, and interaction data in these networks.
Therefore, because she learns, her narratives and character split throughout different networks.
In the past, whenever these competing narratives collide, Aisha’s team responded accordingly to heal the break. But they developed a new component for Quin where she seamlessly integrates the two groups and new connections that might conflict through a content adjustment, often compromising between her two performances.
Every major company uses methods similar to Aisha’s, though calibrated differently. Any news corporation pulls all data available for any reader or potential consumer, and develops language to perfectly target that user with a frame and presentation ideal for whatever their goal. If the goal is to convince a group that an oncoming war is in their favor? Based on their interactions on all social network sites, where they live, who they talk to, how they talk to them, their engagement with media, and so forth, they design their media, deploy them, and get the proper engagement with that media to convince that user the war is a positive. The choice of language, whether it’s text, augmented reality, virtual reality, videos, audio, who is delivering the message, what tone they use, what content they site, what emotions they reference, what imagery do they insight, and beyond.
“But the news’s use of these tools lead to propaganda, manipulation for the state or news source’s needs, misinformation and broken democracy,” Aisha tells herself. “We’re making people’s day better and helping them feel more connected, not actively manipulating in any meaningful way.”
Quin does have relationships with news outlets, however, as Aisha has seen in these maps. She’s learned that in order to thrive in many of these networks, she must engage with the news. For example, Aisha sees on the map new popular keywords emerging across many of Quin’s networks about Roberta, a hurricane. In some, she’s offering advice and kind words of encouragement that all will be ok. In others, she’s condemning those who are in disbelief that the hurricane is real. On Virtua, she’s showing empathy and support to the working man, the vets, and those who feel taken advantage of by the government. “It’s always interesting,” Aisha reflects, “when she takes a bit more of a biting tone.”
It’s an interesting balance. In some networks, breaking-news offers a boost in social capital, and a boost in engagement. We see her trying to find possible hit stories or potential moments of drama and drop them right when the news sites on the network post them. Quin’s helping her bored sub’s have as much fun as she can through debate and discussion.
And it’s not always about breaking the news. Sometimes, it’s more about the critical eye offered about that news. Based on how she interprets the ecosystem around her in a specific network, Quin might judge that breaking the news is not the most important act to improve engagement, but rather being the first to comment critically. Further, in many instances it is not essential to be the first to comment, but rather a commenter a few days after breaking, when the news has begun to settle in, user’s are beginning to reflect on that news and change their tone slightly. In these cases, she performs as a motivational and inspirational figure for those with a more critical eye.
One of Quin’s components Aisha finds most interesting is the way she creates enemies and allies. Depending on the network attributes, she will identify another influencer as a key person to develop a public friendship with. The network often craves kindness, sense of community, comradery, care shown between people they admire. These are often highly strategic, and require a similar analysis from the part of the other influencer. More often than not, they analyze the networks they inhabit and both come to the conclusion that a friendship is strategically in both of their favor. In rare instances of misalignment, the backlash can cause Quin’s social capital to take a hit.
Across other networks, Quin identifies useful enemies and “frenemies.” Through her analytics, she understands that engagement and activity within the network is stimulated by conflict. Yet, she shows us that conflict can manifest in many different ways. In some instances, it is a battle of ideology and deep belief, in which she clashes with another influencer’s belief system they have outlined to their subs explicitly. In these cases, she can choose to ignite this conflict through passive, subtle moves, small shots fired, or sometimes in a dramatic, large, visible, and painful way. It’s an incredible feat of the system to navigate such complex relationships with strategy, and for that Aisha feels deeply proud.
But again, all of this is to foster greater connection and help people who feel lonely, depressed, and isolated, and to help some folks feel entertained, inspired, invigorated. Aisha feels great pride, though sometimes inclingings of doubt. Back at the map, she stares as isolated groups grow closer to one another, and closer, and closer, it seems impossible for them to not allow edges between these nodes of different clusters. But, for some reason they change paths right before colliding. Aisha worries and wonders, “Why don’t they collide?” As she zooms out on the map, it feels as if the networks are getting further isolated. Aisha zooms in on the keywords and language in each. Her gut drops, as she realizes Quin denies the hurricane in one, and condemns those who do nothing to help the most vulnerable in another.
Her pride abruptly transforms into doubt, “Is Quin controlling the structure of this network of these people’s relationships?”
Roberta: U.S. Warns of “Mass Casualty Event” as Conspiracy Theories Swirl
MIAMI (AP), 5:43pm -- Federal and Florida authorities issued dire warnings Wednesday about the expected landfall on Saturday evening of Hurricane Roberta, and about attempts by armed militia groups to interfere with the evacuation of Miami.
Roberta, the 34th named storm of a very active 2048 hurricane season, is by far the most intense hurricane on record. The National Hurricane Center reported sustained wind speeds of more than 250 mph early Wednesday, with gusts to 270 mph, easily clearing the previous record of 224 mph set by Hurricane Joseph in 2044. The NHC attributed Roberta’s unprecedented strength to high water temperatures in the South Atlantic, driven by continuing climate change, and noted that it was not expected to weaken appreciably before landfall. Given Miami’s low elevation and several decades of sea-level rise, forecasts of Roberta’s storm surge at landfall from the NHC and the National Oceanic and Atmospheric Administration exceeded 20 feet.
“This is an apocalyptic storm,” said Gov. Jolene D. Sanchez (R-FL) in a blunt press conference. “The wrath of God is about to hit Miami and you have to evacuate now. Anyone still in the city at landfall must understand that we won’t be able to protect you.”
Since it became clear three days ago that Roberta would make landfall in Miami, Sanchez and a federal task force have been organizing the evacuation. Despite logistical support from the Department of Defense, the effort has been dogged almost from its inception by conspiracy theories and sometimes-armed local opposition.
Rumors circulating on social media, especially the popular Virtua virtual reality app, allege falsely and dangerously that the hurricane is not real, and that the evacuation is an attempt by the government to put the evacuees in labor camps or otherwise infringe constitutional rights. One viral VR immersive noted a task force plan to use Amtrak and freight trains for evacuation, darkly comparing it to Nazi deportations in World War II.
Beyond popular suspicion aroused by the conspiracy theories, armed groups acting on them have become a problem for the evacuation effort. Several left- and right-wing paramilitary or militia organizations are known to be involved. This morning’s bombing of CSX rolling stock north of Miami was a particular blow, with the amount of time required for clean-up sharply reducing rail capacity and pinning the government’s hopes on road transport over I-95.
The FBI issued an urgent warning Thursday afternoon of intelligence suggesting plans by other militia groups to disrupt traffic along I-95, by occupying the road, stranding trucks or even destroying overpasses. FBI Director and evacuation task force member Robert J. Wood implored anyone with information to contact his agency or the task force, warning that “we’re facing a mass-casualty event if 95 isn’t usable.”
The conspiracy movement surrounding the hurricane evacuation “came almost out of nowhere,” according to Wood. “We’ve seen these sorts of conspiracy theories before, but never so quickly or so well organized on such short notice.”
Regardless of the action on social media, the director said, the message to the public is clear: “For God’s sake, get out of Miami.”
Markus was angry, but that was nothing unusual. At 34, he had been angry more or less continuously since leaving the service six years ago. He had drifted through a succession of bad jobs and even worse relationships, moving from place to place in the hope of leaving behind his gnawing rage at how it had ended. Risking death on a sardine can in the ocean wasn’t something Markus had enjoyed, exactly, but he hadn’t spent years doing it just for those traitors in the government to give up and settle the war. Not after they had attacked us first.
But Markus wasn’t the President or an admiral or a senator, and without any ability to do anything about it he had gone back to civilian life on demobilization, just like everyone else. What he hadn’t expected, the silver lining of it all, was how the great betrayal had opened his eyes to the conspiracy. Without it he might have gone through life not knowing.
It was all around him, once Virtua and his favorite virts had shown him where to look: the chemicals in the water, the ultrasonic bugs in mail trucks, the Rothschilds surveilling his bank account. The vaccines they were giving to kids might just be the most galling part, but it wasn’t the most urgent. That honor went to the fake hurricane the government now said was heading toward Miami.
Markus lived in Miami now, and for a change he liked it. When he wasn’t on Virtua, with the headset over his eyes and ears and the transcranial cap on his bald head, there was a lot to do in the city. The sunny beaches, the packed clubs, the food -- they were all good distractions for Markus, but he was especially pleased that he had found a circle of like-minded people who also understood the conspiracy was real. The media might lie about it, but he and his buddies knew the truth. They had different backgrounds and had come to their awakenings in different ways, but the knowledge bound them together.
And yet even they were surprised by the brazenness of the hurricane power grab. Sure, hurricanes were real and you had to evacuate sometimes, but the most powerful hurricane ever? Suddenly heading right for them, complete with apocalyptic warnings about storm surge, only a few months before an election that Florida might decide? Nah. Had to be fake. His last doubts were extinguished by a viral virt about the feds’ plan to cram people into freight trains for the evacuation. Just like it said, in an especially convincing point, the Nazis had done the same thing!
But Markus was also surprised, and heartened, by the incredibly rapid rising against the hurricane fakery that he saw on Virtua. He had never seen anything like it: in just a couple days, an astounding number of people came together to declare that they had had enough. Virts like the one he saw about the trains had sparked it, usually from little-known accounts, but wasn’t that even better? Normal people were done rolling over for the conspiracy.
Markus and his buddies agreed that getting angry wasn’t enough this time: they had to do something about it. Johnny, Markus’s oldest friend in Miami and just as outspoken an opponent of the conspiracy, knew some people on Virtua, harder men than they were, who had read him into their thrown-together plans to stop the trains and then the truck convoys that would inevitably replace them.
Neither of them hesitated when asked to join.
It was quiet in the office at 2am. The conversations and keyboard clacks that could be heard during the day had given way to the hum of air conditioning and the furtive scurrying of the occasional mouse.
Except for the mice, Abe liked it that way: He could be alone to think without getting distracted by meetings that should have been emails. (And the view of the Embarcadero from his office on the 15th floor was even better by night.) He did his best research at 2am, always had, since well before starting at Virtua. And it was that research that had gotten him hired, after all. He was an AI researcher, and as he’d be the first to tell you, he was one of the best. A long string of publications, best paper awards and a citation count that could be mistaken for his net worth had given Abe his pick of jobs. Turning down Harvard had been difficult, but when his prospective boss at Virtua described this job, Abe knew he wouldn’t be able to say no.
Virtua’s business was virtual reality, and the thing about virtual reality was that it could get a bit too real. Millions of people were interacting in something that wasn’t quite real life, but unlike the early prototypes in the 2010s, it was close enough. The possibility of providing input to -- and reading emotions directly from -- the brain through the transcranial caps Virtua had invented made all the difference. Enormous breakthroughs in quantum computing and superconducting electronics in the 2020s and 2030s had been critical too. But while the new VR was widely agreed to be better than the early days of smartphones (and universally agreed to have improved online dating), there were some downsides. After Virtua started its first offline religious war, management had decided to make some changes.
Which is where Abe came in. He had been recruited with the goal of building Virtua a social central planner, an AI that could manage the passions the virtual reality environment stirred up in its users. What that meant had been left deliberately ambiguous, and he had taken an expansive view of it.
Abe looked at his six monitors without really seeing them, lost in thought. The model his team had built was a technical marvel: it could look at the entire Virtua service at once, identify budding social problems, and nip them before they could bloom. Usually the changes this took were small enough the users affected didn’t even notice. What was a sock puppet here or an allegedly random change in simulated weather there if it juiced engagement and got the government off their backs? (Research ethics had never been Abe’s strong suit.) Their competitors, especially Cybex down in Miami, didn’t have anything like it. And yet lately there had been some problems. Signs of a social planner making mistakes.
It must be the hurricane, Abe thought. The wave of insanity forming on Virtua ahead of the Miami hurricane had surprised him, but he figured not everything could be controlled. Some things people felt too strongly about for the AI to be able to tamp them down. Yet.
With that thought, Abe sighed and turned his attention back to the monitors. Somewhere in all this debugging output the answer was waiting for him to find it.
After another fruitless hour Abe was about to call it quits when he noticed something. Several of the AI’s sock puppets had a remarkable pattern of close contact with some unsavory characters in Miami. Aggressive militia members with anger management problems were certainly the sort of people he wanted the AI to keep calm, but it wasn’t doing that. It seemed to be egging them on! With a sinking feeling in his stomach, Abe tried the model’s voice interface.
“What can you tell me about this Markus, in Miami? And his associates? Why are you working them up like this?”
“User Markus is a 34-year-old man in Miami with a recent history of Virtua use and intermittent content violations, requiring active management --”
Suddenly annoyed at how tired he was, Abe realized he had to ask more directly. “No, that’s not what I mean. Identify the motivating supergoal.”
“Profit? What the hell does that mean? How does amping up some domestic terrorists make us a profit?”
Listening to the explanation, Abe very nearly fell out of his chair. “Is this a joke, you goddamn tin can? I don’t remember building a sense of humor into you, but I sure hope you’ve grown one. You’re telling me that,” and here his voice caught on the dark absurdity of it, “you’ve been whipping these conspiracy theories up all along? And you want them to blow the I-95 overpass so that Cybex staff will be stuck in Miami? Did I hear that right?”
“Competing aggressively and maximizing quarterly earnings are core Virtua values.”
Even ethically challenged Abe wanted to gag on that one. “I’ll tell you what’s a core Virtua value, calling the FBI. Call the agent in charge in Miami, right now.”
“I’m sorry Abe, I’m afraid I can’t do that.”
Struck both by the realization that he was living through an AI-alignment nightmare and by how terrible a sense of humor his brainchild had, Abe yanked out his phone only to find that, somehow, it had no signal. Bursting through the doorway, he ran down the hall in a blind panic, looking for the hardwired video call booth. He wasn’t thinking of the FBI anymore, just the engineer on duty at the data center. They had to shut this thing down as fast as possible.
A deeper sense of panic set in when he reached the video call booth and realized none of the equipment would turn on.
Back in his office, after finding that none of the exterior doors would open and the voice interface had stopped responding, Abe slumped back in his chair. Turning it over in his mind, he wondered what separated him from Markus, really. How did he even know the hurricane was real? He had only heard about it through the same electronic devices his malevolent tin god had shown itself so skilled at manipulating.
Casting one more defeated glance through his office’s glass door, he saw Thursday-morning CNN on the common room TV, reporting live from Miami in front of a brilliantly clear dawn, not a cloud in sight.