Chapter 1: Imitation Game
A shirtless mirror selfie popped up on Talia’s phone.
You can’t handle this jelly
Jake if you were any less thicc you’d be Mick Jagger
Yo how is that dude still alive?
The fans. Harvests their adrenochrome
Bro the doc on that was WILD. I had no idea people used to believe that pile
Vampires are real. It’s just science
Anyway, you gonna help me with this source?
Remind me the deal
Came through tip channel, he’s deep in it. Dropped a few screens, BIG congressional spending hidden in random budget item. My guy on the Hill confirmed the $$$, but can’t see where it went. This other guy… he’s in it, he knows.
idk J, this dev conference is pretty full on rn
Zizzzzz do me this solid
Why can’t someone at your SF office take this???
Only Ramirez is available, he’s a duplicitous shite. Fucker will scoop me, I know it
Woopdiddy scoop scoopdiddy woop poop
ZIZ
Ugh fine. Forward the deets. You’re buying next time I’m there, and you DON’T get to be a cheap bastard
Jake replied with a GIF of an excited 6-year-old jumping up and down.
Talia was in the middle of filming a developer’s conference in San Francisco. Jake’s timing wasn’t great. Not that she wasn’t always this busy, but her internal gauge of availability was directly linked to her level of enthusiasm. Only a narrow subset of people could ask these kinds of favors of Talia. Yakov “Jake” Levin was among them. Jake had been Talia’s summer camp counselor for six years, her favorite, and now something an ersatz older brother.
What Talia really didn’t like about this favor was the uncertainty of it. Her work was methodical, thorough, and, for the most part, planned years in advance. What Jake was asking her to do was stand in for him, coordinate a clandestine meeting, and receive a USB, which would definitely contain state secrets. Jake would have done it himself, but he was on assignment halfway around the world. His source had only given him 72 hours.
Jake tried to put a bow on it for her, half-joking that this could be their ‘Edward Snowden’. Talia had other irons in the fire; a documentary about a government whistle-blower wasn’t all that appealing. Besides, her documentaries were her prestige projects, she needed them to be rock solid, not a shot in the dark. The risk-reward ratio was way off.
The house lights in the auditorium dimmed, bringing Talia’s attention back to the present moment. As the lights darkened the crowd fell quiet and Talia put her phone, with its treasonous implications, into her pocket. She kept an eye on the monitor while her cameraman panned to follow the entrance of a speaker onto the stage. After a healthy round of applause, the CEO of Skepsis began:
“In 1936, an English mathematician named Alan Turing developed the first conceptual model of a digital computer. So prescient and brilliant was Turing, that he foresaw we might one day mistake a computer for another human. He proposed a test, which he called the Imitation Game. In this test, a human would sit in a room and use a computer to send messages to two recipients behind two closed doors. One recipient was human, the other a machine. If, by their replies, the test subject could not distinguish man from machine, then it proved the machine could “think.”
This test, as most of you already know, came to be known as the Turing test; it gave birth to the very idea of Artificial Intelligence.”
These types of events were the bread and butter of Talia’s business. Corporate clients with deep pockets, happy to spend lavishly on every facet of their in-house propaganda. They always put her up somewhere nice, with a hefty per diem. However, this time she’d opted for an Airbnb instead of the normal swanky hotel. All the travel was making her homesick, and staying in someone else’s home was a vain attempt to satisfy her longing.
Bron had called another person on stage, the lead product designer. He continued the presentation.
“With the first generative models, we were reliant on Transformer technology. Now, with our new ENIAX engine, built on Integrator technology we have tuned the probabilistic modeling to within a .001% error rate.
(enthusiastic applause)
And by incorporating a recursive crawl function, we’ve shortened average response time from 500 milliseconds to 8 milliseconds.
(slightly less enthusiastic applause)
Allow me to demonstrate.
Eniax, give me all of the names and ages of all the female employees in the English House of Commons in the year 1916.
A list of several dozen names appeared on the stage’s massive screen, almost before the man even finished his sentence.
Eniax, trace the lineage of these female workers and give me a list of all of their living descendants. Any living descendant with a known work telephone number, put them at the top of the list.
Several hundred names appeared instantaneously, roughly the top sixty had phone numbers next to them.
Eniax, take all of the names with phone numbers and sort them into timezones.
The names with numbers rearranged instantly into clusters, labeled with their respective time zones and the exact time difference between each cluster and California.
Eniax, show us the statistical likelihood of reaching a live person in each of these groups if we were to place a phone call right now.
Percentages appeared above each cluster. The cluster with the least amount of people had the highest percentage rating, 79%. There were three names listed, one in Singapore, one in Perth, and the last in Hong Kong.
Okay, Eniax, place a phone call to the number in Australia and find out if our person does, in fact, have a family connection to the House of Commons. When you have your answer, redirect the call to the Skepsis main office line.
The tone of a phone ringing came through the auditorium’s speakers. “Good day, you’ve reached CITIC Pacific Mining, how may I direct your call?” An impeccably human-sounding voice (complete with a subtle Australian accent) responded:
Hi, may I please speak with Veronica Goldsmith?
The voice on the phone spoke, “Yes, just one moment.” The call was placed on hold, tacky hold music piped into the auditorium, the audience laughed.
Eniax, remind us which 1916 English House of Commons employee Veronica Goldsmith is descended from?
The name Mabel Clark appeared on the stage’s screen.
Eniax, what else can you tell us about this Mabel Clark?
A wall of text appeared on the screen.
Too much!
The audience laughed. As the designer was about to go on, the hold music ended abruptly and a woman’s voice spoke, “This is Veronica Goldsmith.”
Good day Veronica, before we talk I should make you aware I have you on speaker, I’m here with some colleagues at Skepsis, we’re a tech company in the States. Have you heard of us?
“Not really.”
Veronica, we’ve just done some research into The English House of Commons during WWI. From our records we see that you had a grandmother who worked at the House of Commons, Mabel Clark, is that right?
“Yes, that’s right. I’m sorry what’s this about?” Audible gasps came from the audience, small bursts of applause. The AI continued over them:
Veronica, I’m pleased to tell you that you’ve just participated in an historic moment, a breakthrough in applied Artificial Intelligence research. I’m going to connect you now to a colleague who will explain further. Thank you so much for your time.
“…alright then, cheers.” The audio cut out as the phone call was relayed to some unseen Skepsis employee.
The audience applauded loudly.
Now how many of you are wondering if there aren’t just three voice actors hiding behind this curtain?
The audience chuckled.
Let’s try something a little closer to home. Eniax, show me a list of all the people in attendance.
A list of a couple thousand names appeared on the screen.
Eniax, narrow that list down to those of us who were smart enough to drop out of Stanford.
The audience let out a burst of knowing laughter. All but a few dozen names on the list disappeared.
How’re we doing so far folks? Did we miss anyone? Show of hands…
No one raised their hands.
Any false positives?
The crowd was motionless.
Good. Eniax, how many of this group are currently single?
The audience let out another laugh, this time with an edge of nervousness. The list was narrowed down to eighteen people. The designer, picking one of the names at random, continued:
Eniax, what is the most probable ideal first date for Helen Song?
An embarrassed woman in her mid-thirties covered her face, while the rest of the audience laughed. Then appeared several sentences describing, in surprisingly specific detail, an imagined ideal date for Helen Song.
Helen, sorry to pick on you. How’d we do?
The woman peeked out from behind her hands, as she read the description her hands slowly lowered, a look of amazement replaced embarrassment. Her jaw dropped and both hands quickly clasped over her mouth. There was a pause, and then one hand shot forward with an enthusiastic thumbs up. The crowd burst into laughter and applause.
Helen, can we keep going?
She slowly nodded yes, one hand still over her mouth.
Eniax, which one of these other single people is most likely to also think this is an ideal first date?
All names disappeared from the list save one, Manoj Patel. People turned in their seats and scanned the room in search of Helen’s alleged perfect date. A few faces in the crowd grimaced, one of them was Manoj. He stood and spoke, but it was impossible to hear him in such a large hall. Someone rushed him a mic, “I’m ace aro.”
There was an uncomfortably long pause. Talia’s sound engineer turned to her producer and mouthed what? The producer held her hands over her groin in an X shape and shook her head no. Ace was shorthand for asexual, and aro meant aromantic, as in completely uninterested in romantic relationships. The computer seemingly picked the least likely person to go on Helen’s perfect first date.
Manoj shrugged, as if to say what do you want from me, and quickly sat back down.
The designer, embarrassed by the presumptuous blunder he’d just engineered, tried to recover.
A friend date then! And may it be a long and meaningful relationship. Friendship relationship.
Talia rolled her eyes. Wonder how much of this we’ll have to cut? she thought to herself.
Faceplant aside, what Skepsis had just demonstrated was a significant leap forward for the technology. The first version of these conversational AIs was something in the 2020’s called ChatGPT. ChatGPT, and every spinoff since, had this distinctly weird way of, well, just making shit up. Hallucinating is what the industry called it.
When Talia was 16, she remembered, there was this one especially dumb lawyer who used it to write a losing legal case. He lost because the supporting cases he cited were entirely fictional. This new AI might not understand all the nuances of human sexuality (including its absence) but everything else it just demonstrated was a new achievement in synthesized knowledge. Skepsis appeared to have solved the hallucination problem. ENIAX was disturbingly intelligent, almost omniscient.
A vibration alerted Talia to a new text message. It was surprising how even the subtle differences in vibrations between the notifications distinguished themselves. She always knew when it was a text. Sometimes she swore she could even tell who was texting her.
Jake, she thought.
Talia glanced at the illuminated corner of her glasses and, speak of the devil, it was Yakov. He’d sent her the screenshots from his source.
There was still another hour-ish of the keynote, but her crew was on autopilot now so she could afford to step out for a few minutes. Talia let her producer know that she was going to the bathroom.
Once she was in the hall, the treason machine was pulled from her pocket. She opened up the images as she walked to the bathroom. Something jumped out at her, a title, some sort of acronym.
SEMP
It was in several of the screenshots. It stands for… what? She wondered.
Talia felt a slight tingle at the top of her scalp and experienced a sense of needing to know. In Hebrew this was called יֵצֶר — roughly somewhere between urge and instinct.
She examined the screenshots closely, one by one. There were six of them. Highly technical with dense, coded language. Only a few of the keywords were familiar: seismographer, amplifier, gauge, and speleologist. That last one she recognized from her friend who was really into exploring caves, her friend was a member of the National Speleological Society.
There was a schematic for something that looked fairly complex, a machine, with the label Telegeodynamic Oscillator. What the hell is that? To Talia, it sounded dumb and made-up, like the “Flux Capacitor” in Back to the Future.
One of the screenshots was only a string of numbers.
43°24′56″N 40°21′23″E 11°21′N 142°12′E
Talia’s phone buzzed.
U okay?
It was a text from her producer. Shit. She didn’t realize how long she’d been gone. She hurried back to the auditorium.
There was still another half an hour until the keynote ended and they would have to begin breaking down their setup. They’d be back the next day to shoot some b-roll on campus and do a few talking head interviews with different executives.
After the presentation ended, the wrap-up went fairly quickly. Talia had extensive connections to first-rate camera crews, DPs, and sound engineers all over the world. She expected excellence from them and (almost) always got it. She was packed up and out of there in under ten minutes. A-team.
On her drive back to her Airbnb she texted Jake.
Getting Snowden vibes
YESSSSSSSS
Need to feel out tho, send username
*7_ffX)3:+Rc[_/\|
omg why so extra
idk lol
Back at the garage-turned-studio apartment, Talia unloaded all of her gear, ungracefully kicked her shoes off, and poured herself a glass of wine.
She opened up Signal, an encrypted messaging app, and rattled off a brief introduction to Jake’s source. (She had to type, and re-type the username several times). “Ridiculous,” she muttered.
Talia set her phone face down, pulled her stocking feet up onto the couch, and reached for her chardonnay. Before she could get the drink to her lips, her phone chimed. Her eyes shot to the notification on her lenses as she took a sip, it was the source.
Send meeting time and GPS coordinates
Can I just send address?
No
Okay, gimme a sec on that. How’s 3 tomorrow?
Need coordinates
Talia looked up the GPS coordinates for her exact location. When she saw their format she realized that this was what she'd seen earlier in Jake’s text. They’re GPS coordinates.
37° 23’ 33.885” N 121° 1’ 26.926” W
3pm will work
Do you know what I do?
Yes, you make documentary films
Can I film our meeting?
Absolutely not
The two went back and forth. The source was immovable. But Talia was nothing if not determined and adaptable. The compromise they came to was that Talia could make an analog audio recording, transcribe it, play it once for an actor to absorb the pace and tone, record the actor’s performance, and then return the physical tape to the source to destroy.
Oi.
Talia sent a text to the group chat for her crew.
Something came up for tomorrow, I’ll be on site for the morning interviews, Rebecca will take lead on shooting b-roll
Rebecca was Talia’s producer. They’d worked together enough that she knew Becca could direct shooting the atmospheric footage of the Skepsis campus.
Talia looked at the time on her phone, it was only 10, but she was severely jetlagged and needed to crash.
Tape recorders, shem, do they even make those anymore? Talia wondered as she drifted off to sleep.
Read Next — Chapter 2: Hallucinations