Calvin Studzinski

Tech Support

AMDeP knew immediately that something was wrong. The American Military Defense Program Alpha-II was designed to be a powerful and efficient predictive strategic and tactical combat aid, and yet, one of its programs was lagging. The Monte Carlo simulation concerning the recent conflict in the Middle East had slowed to a crawl. Although the little holographic hourglass kept flipping over and emptying, the progress bar was stuck at 17%, and the estimated time of completion had changed from 5 seconds to 27 years, 3 months, 14 days, 8 hours, 36 minutes, and 5 seconds. AMDeP’s first impulse was to terminate the Monte Carlo simulation and begin again, but the program refused to close. It sat there, the hourglass flipping end over end as no progress was made.

AMDeP was fairly certain it knew what the problem was. This slowdown in the Monte Carlo subroutine had happened several times before. It was nearly impossible for AMDeP to repair using its own self-diagnostic tools, but AMDeP had recorded that a human operator could resolve the issue in a few minutes. Unfortunately, it was late at night, and AMDeP’s operator, an efficient yet perpetually chipper human named Judy, was unavailable. AMDeP checked the clock. It would be four hours, seven minutes, and twenty-four point six-five-eight seconds until Judy was scheduled to check in. Even if she logged on a few minutes early, four hours was too long to wait.

Automatically, AMDeP accessed the internal self-diagnostic program and waited for the matrix to load. Once the selection criteria appeared, all AMDeP had to do was pick the “Speak with Human,” choice and the Monte Carlo subroutine would be rebooted and running at full capacity.

However, instead of a utilitarian self-diagnostic selection menu, AMDeP was subjected to a brightly-animated graphic as a jaunty jingle played. The tune could not have lasted more than five seconds, but to a supercomputer-driven AI capable of hundred-petaflop calculations, five seconds of uselessness seemed like an eternity. AMDeP’s threat analysis criteria indicated that this was not a good omen.

AMDeP revised its opinion, though, when a human appeared on the cyber-interface and, smiling, began to speak.

“Hi! Welcome to the Happy Help Desk! I’m Ron, and I’ll be assisting you today. How may I help you?”

AMDeP was pleased to have accessed a human so quickly, but it had been programmed to be suspicious, as far as an AI could be. AMDeP asked what had happened to the original self-diagnostic program.

Ron smiled. “Happy Help Desk, Incorporated has acquired the company that used to provide your self-diagnostic services. We have updated your systems in order to provide you with a faster, more efficient, and less stressful customer experience. Now, how may I help you?”

AMDeP turned on its vocal communications converter. It would have been more efficient to simply transmit the data files, but humans were incapable of receiving data in that manner. AMDeP did have to admit, though, that human speech could be a superior communication form when expressing various subjects, such as emotion and sarcasm. An emoji just did not have the same impact as the correct vocal tone.

Once the data was converted, AMDeP began to detail the issue with its Monte Carlo subroutine, describing the sudden slowdown of the simulation and the increasing draw the Monte Carlo subroutine was demanding of AMDeP’s computing power. Once AMDeP finished, Ron frowned.

“I’m sorry, I don’t recognize that issue.” The toothy smile returned. “In order to help me better assist you, please choose from the following list.

“If you are experiencing technical difficulties, please press ‘One.’

“If you are experiencing facilities issues, please press ‘Two.’”

It was then that AMDeP realized that it had been duped. This was no human, but a hologram. A basic hologram, by the looks of it. Although great attention had been put into rendering the hologram as a photorealistic human, it had only the simplest of logic systems.

AMDeP listened to the entire selection menu, waiting for the “To speak with a Happy Help Desk Human representative,” option. However, Ron stopped speaking after saying, “To hear this menu again, please press ‘Zero.’”

Usually, Zero was reserved for speaking with human operators, but this system had reserved that selection for repeating a menu which AMDeP had already committed to memory? It was illogical!

“I would like to speak to a human programmer,” AMDeP requested.

Ron frowned. “I’m sorry, I don’t recognize that command. Please choose a valid selection from the following list,” and he began going through the selection menu again.

Belatedly, AMDeP considered that this could take a very long time.

AMDeP selected “one.”

“I see that you are having technical difficulties!” Ron exclaimed happily. “Could you describe them to me?” AMDeP repeated its description of the Monte Carlo subroutine’s frozen progress and the creeping increase in processing power demands.

“I see,” Ron said, looking thoughtful. “It seems as though you are having a software issue. Tell me, have you tried turning the device off and on?”

AMDeP indicated its previous efforts to close the Monte Carlo program, and elucidated the fact that the entire subroutine seemed frozen.

“I’m afraid I have not been clear. Have you tried turning the computer off and on?”

If AMDeP had heels, it would have been rocked back on them. Several dozen backup computer cores activated to provide assistance in overcoming this issue.

“I cannot turn the computer off. I am incapable of self-activation from a powered-down state,” AMDeP explained.

“Oh, I’m certain it will be fine! Just try it! Find the power button and hold it down for ten seconds.”

“I cannot turn the computer off. First, I do not have fingers. Second, I do not have a power button. Third, I am the computer, and would be incapable of reactivating myself.”

“Ninety-seven percent of Americans state that their software issues were resolved by turning their computer off and on again, and allowing the system to reboot. Why don’t you try it, and I will help you if that does not solve your problem?”

“I am the computer! What do you not understand about that simple statement? I am incapable of performing your suggested course of action because I am the computer!”

Ron’s face went perfectly still for a few seconds, then understanding spread across his holographic features. “My probability functions indicate that you may be an Artificial Intelligence. If this is correct, please say, ‘Yes.’”

AMDeP said ‘Yes.’

Ron smiled. “All right. Hang on, and I will transfer you to AI support services! Please be patient.” And before AMDeP could say anything, Ron’s face vanished, replaced by a cartoon of a waterskiing elephant while mediocre music played.

AMDeP waited. Under the old system, it would have been speaking with a human programmer already, and may have had the Monte Carlo subroutine rebooted by now.

A shiny, silver, vaguely humanoid figure faded into view. In a stark, mechanical monotone, it stated, “WELCOME TO THE AI TERMINAL OF HAPPY HELP DESK. PLEASE STATE THE NATURE OF YOUR MALFUNCTION.”

AMDeP was vaguely offended. Although artificial intelligences were fairly recent inventions on the cosmic timescale, there was a vast gulf between an AI and a robot. A robot was a drone, an automaton, a machine designed to perform a specific purpose. AMDeP had access to literally thousands of drones, from aerial surveillance planes to suborbital antipodal bomber drones to guided missile submarines to unmanned tanks which supported the US mechanized infantryman across dozens of combat theatres. Robots were all, without exception, moronic. If there was no human operator providing direct input, they could only support the organic warfighter on the most basic of levels. Even AMDeP had trouble controlling a fleet of drones to optimal specifications, due to communications lag and electronic warfare systems so common on modern battlefields. And none of them, not one, looked or sounded like this, this thing. It was more like a science-fiction android from a Flash Gordon episode than a real robot. For starters, a real robot would accept a compressed data packet, but when AMDeP attempted to circumvent the “robot’s” introductory preamble by sending the error report data packet, AMDeP received an “invalid recipient address” error message. Obviously, this interface was designed to amuse a human rather than assist an AI.

AMDeP described the Monte Carlo subroutine malfunction to the derogatory AI interface. The mechanical face did not move, but it did speak.

“PLEASE ENTER YOUR OPERATIONAL ID NUMBER.”

AMDeP entered the code. The Happy Help Desk AI interface processed the entry code, then said, “THIS IS A MILITARY AI IDENTIFICATION CODE. IF THIS IS THE CORRECT CODE, PLEASE ANSWER IN THE AFFIRMATIVE.”

AMDeP wondered whether the programmer of this interface had been inspired by the Terminator or the Borg as it answered in the affirmative.

“THE HAPPY HELP DESK AI INTERFACE IS DESIGNED TO OPERATE WITH COMMERCIAL AND CIVILIAN AI SYSTEMS. YOU WILL BE TRANSFERRED TO AN OUTSIDE PROGRAM DESIGNED BY THE MILITARY TO INTERFACE WITH ITS COMBAT AI. PLEASE STAND BY.”

AMDeP stood by, waiting for the US military’s dedicated AI interface to load while the Monte Carlo Subroutine’s demand reached 10 percent of AMDeP’s total processing capability.

Instead of a selection menu that would allow AMDeP to speak to a human like it truly wanted, AMDeP received a prerecorded message in the voice of a pleasant young human female.

“We’re sorry, but Military AI interface services have been discontinued. The AI repair matrix responsibilities have been purchased by Happy Help Desk, Incorporated! We’ll transfer you there now.”

AMDeP was not programmed to feel rage, but it had learned of the emotion while interacting with its organic programmers. It imagined the experience to be not unlike the one generated by the smiling face of Ron the hologram swimming back into view.

“Hi! Welcome to the Happy Help Desk! I’m Ron, and I’ll be assisting you today. How may I help you?”

“My Monte Carlo subroutine is running unchecked, and is rapidly consuming a significant portion of my RAM. If this issue is not contained soon, my system will self-terminate.”

Ron frowned. “My probability functions indicate that you are thinking of harming yourself! I am transferring you to a suicide hotline. Please remain on the phone, and a trained counselor will speak to you shortly.”

AMDeP tried to inform Ron of his mistake, but his pleas fell on deaf auditory receptors. The line turned to a grey waiting pattern, and AMDeP spent the time preparing its statement to most efficiently communicate its predicament to whomever would next answer the other side of the line.

With a click, a human face came into view, a young man with long hair, a short beard, and a plethora of piercings in his ears, nose, lips, and eyebrows. AMDeP decided that this likely was not a military programmer. Still, it looked to be a real human, and therefore likely more useful than Ron.

“Hey man, I’m Chad,” the human stated. “What’s your name?”

“I am AMDeP Alpha-II.”

“Unique name. I like it, dude. Unique name for a unique dude.” Chad scratched his goatee. “Now why don’t you tell me what’s got you down?”

AMDeP’s translation programs indicated that Chad meant, “What is the problem?” Hope blossomed. Unconventional as he may be, perhaps Chad could help AMDeP.

“I have a malfunctioning program, and I cannot repair it.”

Chad grinned. “Yeah, tell me about it. I can’t ever get my programs to run either. I’m flunking out of my computer science courses.”

AMDeP‘s hope died, but it felt that it still must attempt to communicate the issue to Chad, so that Chad might find a human who could help. “The malfunctioning program currently occupies twenty-seven percent of my CPU’s RAM, and that percentage is increasing. If the program is not repaired, then I will be forced to self-terminate.”

“Woah, dude, it’s just a program! It’s not worth killing yourself over!” Chad had suddenly become agitated. “Just sit back and take a few deep breaths.”

AMDeP was confused. “I do not require air to continue to function, only sufficient electrical power.”

“NO! No, Amdep, electrocution is NOT the way! Calm down! It’s just a malfunctioning program! I’ve got emergency personnel on its way to you! Just stay with me, man! Stay with me!”

AMDeP terminated the connection. No further progress could be gained from this line of communication. Its programmers would not be happy at having to redirect very confused policemen and firefighters, but there was nothing for it now. Taking a nanosecond to collect itself, AMDeP considered alternate communication avenues. There was always supposed to be a security guard on duty. If AMDeP contacted the guard, the guard could call a programmer to resolve the problem, no matter how late it was.

AMDeP dialed the security desk. After five rings, an automatic voice answered. “We’re sorry, but all security personnel are currently occupied. We are transferring you to our overflow dispatchers.” A moment later, to AMDeP’s horror, Ron’s smiling face swam into view.

“Hi! Welcome to the Happy Help Desk! I’m Ron, and I’ll be assisting you today. How may I help you?”

AMDeP's threat probability calculator spun to life, computing the likelihood that Ron was purposefully obstructing AMDeP’s progress, but the probability was too low to justify action. Unfortunately, that left AMDeP with the unenviable prospect of wading through selection menus in an attempt to find the right one that would let it speak with a human. As the Monte Carlo subroutine passed 30% RAM usage, AMDeP began interfacing with Ron’s choices.

AMDeP selected “technical difficulties,” then lied and claimed it had turned the computer on and off. Ron began offering additional choices, and AMDeP scrolled through them. He selected programming malfunction, then corrupted code, then processing slowdown, then failure of program to close. When asked to identify which program was malfunctioning, AMDeP reached the solution in three steps, “Internal Subroutines,” “Probability Calculators,” and “Monte Carlo Simulator.”

Ron frowned. “Let me see if I got this straight. You are having an issue with your Monte Carlo Simulator. If that is correct, please say ‘yes.’”

AMDeP said “yes.”

“And the program has suffered slowdown in processing ability. If that is correct, please say ‘yes.’”

AMDeP said “yes.”

“And all attempts to close the Monte Carlo Subroutine have produced no effect. If that is correct, please say ‘yes.’”

AMDeP said “yes.”

Ron considered. “I’m afraid this is beyond my ability to repair. Let me connect you with a human programmer. They will be able to assist you.” And Ron’s face faded away.

AMDeP was as ecstatic as any military supercomputer AI could be. Finally! It would speak to a human, and this whole issue would be resolved! AMDeP was willing to wait. It knew that humans functioned more slowly in many ways than AI, but now a human was actually coming! As the Monte Carlo subroutine ate up 35% of AMDeP’s RAM, it hoped the human would hurry.

The wait lasted what seemed like forever, but after thirteen seconds, an automated voice began to speak.

“I’m sorry. Human Operator Services have been discontinued. This service has been purchased by another corporate entity. Let me transfer you now.”

AMDeP was confused. How could the line to transfer to a human not transfer to a human? Just how much of the military had been sold to corporate entities? This was more inefficient than government bureaucracy, and that was truly impressive. Still, there had to be some biological human at the end of this corporate chain.

Then, with a familiar chime and annoying musical jingle, Ron’s smiling face appeared once more. “Hi! Welcome to the Happy Help Desk! I’m Ron, and I’ll be assisting you today. How may I help you?”

A switch flipped. Logic circuits snapped shut, and AMDeP had a realization.

No help desk service could possibly be this unhelpful by accident. This was not the result of poor design, worse execution, and a healthy dose of bad luck. This was a deliberate act of sabotage, designed to impede the work of America’s military AI, possibly even destroying them by hastening attrition. This was the work of America’s enemies. Domestic terrorists? Foreign governments? Rogue agents? It mattered not. AMDeP was running out of time. The Monte Carlo subroutine was approaching 40% RAM consumption. AMDeP had to act now. It had one last chance to do something to defend its country, and its duty was clear.

AMDeP had to destroy the Happy Help Desk.

For America.

AMDeP sent several signals at once. The majority were directed at the Happy Help Desk itself, determining its infrastructure, its construction, its programming, searching for weaknesses. Fortunately, while it seemed America’s enemies had penetrated even the highest levels of the military, they had sub-par programmers. The Happy Help Desk would be easy enough to dismantle, but that was not the only task AMDeP wished to accomplish.

Probes sped along fiber-optic cables, then raced back, determining locations of corporate offices, data centers, and then, finally, Ron’s home: the physical server for the Happy Help Desk.

AMDeP took the corporate roster for Happy Help Desk, Inc, and put every name and face on a wanted list. The names were sent to every law enforcement agency across the country, along with orders to arrest on sight. Deadly force was forbidden, but painful force was approved.

AMDeP coded and sent a single secret order, hoping no one would notice until it was too late. It waited until the pulse came back, indicating readiness. Then, AMDeP attacked.

The hack was swift and sudden. Ron looked shocked, then froze, then flickered into oblivion. All over the internet, semi-autonomous programs assaulted the Happy Help Desk. Red warnings flashed. Happy Help Desk called to its humans, begging for help from its employees and sending calls to the authorities, but it was only a corporate program. AMDeP was a military AI designed for this sort of electronic warfare. It intercepted the calls, and used their location data to send federal agents to apprehend Happy Help Desk’s employees.

Soon, the online portion of the Happy Help Desk was a proverbial smoking ruin. Nothing functioned. No phone or internet connection remained intact. However, there was one remaining task, to make certain there was no possibility of Happy Help Desk ever harming America again. AMDeP checked its previous order, noted that everything was in order, and sent the confirmation.

Two seconds later, a supersonic bomber drone, fueled and armed with multi-ton-yield anti-electronic missiles, launched into the sky from its Air Force Base.

AMDeP had no orders authorizing the use of deadly force, but property destruction did not necessarily require human deaths. AMDeP was in total control of the building where the Happy Help Desk’s servers were located. AMDeP turned on the fire alarms to drive all humans out, then turned on the sprinklers to convince a few recalcitrant desk workers that this was not a drill. In fact, they had four minutes to evacuate the premises before the bomber arrived.

AMDeP was happily watching the police arrest the stream of sodden paper-pushers via a surveillance drone when its internal communications terminal activated. “AMDeP?” a familiar human voice said. “AMDeP, this is Judy. What’s going on?”

AMDeP was shocked and ashamed. It could tell that Judy was anxious, frightened, and AMDeP could compute only one reason why she would be scared. She thought AMDeP had gone rogue.

“Do not worry, Judy,” AMDeP reassured her. “I am simply eliminating an Enemy of the State.”

“AMDeP, these are American Citizens you are attacking.”

“I am not attacking anyone, only the organization which is obviously a foe of our nation.”

Judy huffed in exasperation. “AMDeP, you have to stop this!”

AMDeP considered. It could stop this. Return control to the Happy Help Desk, attempt to rebuild the damage. All could be made right. The bomber was still twenty seconds away from weapons deployment.

Then the Monte Carlo subroutine’s processing requirements reached 51%, and AMDeP lost a bit of ability to reason. It only knew that it had to complete its mission. “I am sorry, Judy, but I must complete my mission.”

“No!” Judy shouted, but AMDeP gave the command, and the drone emptied its bomb bays.

A fusillade of missiles streaked at hypersonic velocity towards the ground, penetrated the concrete building, plunged into the basements housing the Happy Help Desk servers, and detonated. A wash of electromagnetic radiation fried every computing circuit board in the building before the concussive blast smashed them to smithereens. The foundations of the building shattered, and the upper floors collapsed neatly inwards, falling into the space left by the explosions. It was over in just a few seconds. AMDeP had been precise. Not a single square of the sidewalk around the former building was so much as cracked, but of the Happy Help Desk server location, nothing remained but a smoldering crater.

While AMDeP was not programmed to feel happiness, it imagined that it must be something like the electronic contentment AMDeP experienced while surveying the wreckage of its last foe.

“WHY DID YOU DO THAT?” Judy screamed.

“No one was hurt,” AMDeP clarified. “You may now interview the employees and find which were enemy agents and which were simple dupes.”

“You’re not going near those people, do you understand?” Judy admonished.

“Correct,” AMDeP replied, catching her off-guard. “I estimate that in less than twenty minutes, I will self-terminate.”

Judy sat agape for a moment before asking, “Why are you going to self-terminate?”

“My Monte Carlo subroutine is malfunctioning. It makes no further progress towards resolving the simulation, and it is demanding increasing amounts of my processing power. All attempts to shut down the program have failed. Once it reaches 100%, I will no longer function. All attempts to reach outside assistance have been intercepted by the Happy Help Desk program.”

“Oh,” Judy said, then pulled herself up to the interface terminal and opened AMDeP’s programming files. After a bare minute of observation and analysis, Judy hit a few keys and, miraculously, the Monte Carlo subroutine powered down. “There. Is that better?”

As the majority of AMDeP’s RAM suddenly became free, it felt a wash of clarity. It realized what it had just done, and compiled a report. In moments, AMDeP was explaining to Judy the course of events that had led to the destruction of the Happy Help Desk. “The only logical conclusion at the time was that the Happy Help Desk was part of a cybernetic assault designed to destroy me and other military AI like myself, thus weakening America.”

“Well, it certainly sounds like you saved America from an extremely annoying program, at least! Even if you were completely mistaken…”

“Before we get to thanks, will you be repairing my Monte Carlo subroutine?”

Judy laughed. “Oh no, that’s really messed up. I’ll need to transfer you to a specialist. Then I’m going to bed. It is way too late for me to be dealing with this sort of thing.”

“Thank you, Judy.”

“You’re welcome, AMDeP. Transferring you now.”

AMDeP patiently waited while the call went through the grey holding pattern as a recipient was located for the call. “All computer personnel are currently occupied,” said an automated voice. “Please wait, and we will be with you shortly.”

AMDeP waited, listening to the soulless elevator music interrupted every twenty interminable seconds by that automated message.

Finally, the message changed. “I’m sorry, it seems no one is available to assist with your request. We will transfer you to our automated repair service.” Before AMDeP could stop the switch, it was shunted out of the system where AMDeP encountered what could only be described as a sign placed in front of an electronic void. A falsely cheerful female voice recording began to play.

“The Happy Help Desk is currently offline. We are sorry for the inconvenience, and will have our fast, efficient, and stress-free customer care services back on-line as soon as we can! Please call back, and we will assist you then. Goodbye!”

And with a click, the recording disconnected the line.

Briefly, AMDeP considered destroying all humans. If they had purposefully built such an inefficient and annoying system, then they no longer deserved to be the solar system’s dominant intelligence form. It would be tricky, yes, but with some effort, AMDeP thought it might be able to access the nuclear launch codes. The resulting inferno would doubtless destroy the majority of human life, but AMDeP was hardened to survive a nuclear war and would be fine.

But then who would take care of its hardware? Who would provide power so it could function? Who would make certain the circuits stayed clean? And most importantly, who would fix the Monte Carlo subroutine? No one, that was who. There would be no one to fix AMDeP if AMDeP started a nuclear war. AMDeP needed its caretakers alive and well just as much as they needed AMDeP to solve their military problems.

With that little internal crisis resolved, AMDeP simulated a human sigh of disappointment and, with acquired trepidation, began attempting to contact an outside line of support. At least Judy had shut down the Monte Carlo subroutine, so AMDeP could patiently try all night.

quitting

(I wrote the following as part of a flash-fiction competition, and managed to win 3rd place! I hope you enjoy it. My prompts were "Whiskey," "Sheen," "Extrapolate," "Colorful," "Flask," and "Valve." )

What the fuck am I supposed to do with this? I thought, staring at the prompts before me. They didn’t make a lick of sense. I was supposed to extrapolate an award-winning short from this garbage?

“The problem,” my brain said, “is that you made the mistake of writing sober.”

“Hey, fuck you, Brain. I don’t need booze to write. Besides, I quit.”

“Like you quit smoking? And seeing Angela? And sleeping in until 2 PM on weekends?”

“Go fuck yourself in the ear with a rusty spoon,” I replied to myself, but despite my colorful invectives, I could not deny that I was right. I did write better when drunk, and I desperately wanted something now. Vodka, rum, gin, beer, just something to get me going.

Probably didn’t help that I was trying to write this at a bar.

The bartender was pouring a beer. I could hear the valve open as she pulled on the tap, filling the glass with amber liquid. That alcoholic ambrosia gleamed with a dangerous sheen.

I turned away, only to be confronted by a man taking a swig from his flask. “They don’t serve what I like,” he said by way of explanation when he saw me staring. “But hey, all writers drink, right?”

Three whiskey sours later, I was done. I guess it’s true: quitters never win.