Translated by Marissa Skeels
My job is to follow a set routine, working on a factory line.
I take the same train to my workplace at the same time every morning, do the same tasks, and leave at the same time every afternoon. Lately, business has been so slow that while I show up every day, there’s very little for me to do.
Nothing new ever happens, and I find nothing funny nor sad. It’s like I’m a robot. No, I think it’d be better if I actually were.
K, the new hire, stood next to me. The same guy as always.
“Yo. Did you hear what they said on TV yesterday?”
“They’ve come up with a new, humanoid robot that’s cheap and smart, and easier for factories to take on. They say it’ll put humans out of a job.”
Apparently, they’d also said that it still costs a lot at present, so wouldn’t be rolled out anytime soon. K’s words shook me.
“Pretty soon even we might get a new robot here,” he added.
“Don’t even joke about that. We’d lose our jobs.”
“Could you go anywhere else if you left?”
“Nah. I didn’t get into the right schools, there’s no way I’d be qualified.”
K laughed. “That’s how it goes. No chance a robot would go to school. Yeah, a yesteryear model like you is expensive to run, your cognition’s riddled with bugs, and to top it off you’re a piece of junk that thinks it’s human.”
My job is to sing in a fashionable bar.
I match the beat of percussion instruments, follow the melody of a piano, and sing in a crystal-clear voice. Today’s setlist is all gentle ballads. I stand onstage, clad in a chic dress.
Not many people come on days when I perform. It gets a bit boring, and loneliness inevitably bleeds into my voice.
Unusually, the bar was booked out today for me to sing for a group of customers. Partway through the set, a beautifully dressed woman spoke to me.
“That was a gorgeous, heart-rending song. And I have to say, of all the voices I’ve ever heard, yours is the finest. Would you be so kind as to sing one more for us?”
“Sure, I’d be delighted. I’ll sing for you as best I can.”
“Thanks. Whenever you’re ready.”
No one had ever said something like that to me before. For the first time, I was glad to be singing there.
When I finished, applause rang out. I turned to the lady who’d addressed me earlier. “How was that, did you like it?”
“It truly is a lovely song. I suspect, though, it’d be hard for the average person to sing.”
“Yes, I had to practise a great deal before I could perform it well.”
“Exactly, it’s a trial for humans. Having analyzed how you just sang it, now we can create a voice which will sing whenever anyone likes.”
My job is to follow a set routine, working on a factory line.
I repeat the tasks I’m assigned over and over, quit exactly on time, go back to a warehouse to lie down, and set my thinking program to standby. I guess business has been going well lately, but I’m depressed by just how monotonous the bulk of my workload is. No matter how depressed a robot gets, however, they keep going as they’re programmed to. That’s what it is to be a robot.
The same robot as always, Number 3, stood beside me. Its speech synthesis software gives it its own unique voice. The change of pace I get from chatting with it is all that ever takes the edge off my depression.
“How’re you feeling today?” it asked.
“Robots never feel anything. And we’re not allowed to talk during work, while our programs are running.”
“Uptight as always, huh?”
Perhaps as a whim on the part of people who write their thinking programs, variables are coded into robots which give them human-like personalities. Number 3’s is too friendly. Of course, it carries out its proscribed tasks as it ought to.
“Anyway, about humans,” said Number 3. “Seems like they can adapt to any role you give them.”
“What’s that got to do with anything?”
“When you split a batch of ordinary people up into ‘guards’ and ‘prisoners’ and make them live together, more often than not they begin conforming to the roles they were given.”
“That really is stupid. For humans to be like that.”
“So then, what about you? Working with robots, aren’t you just like a robot that can’t do anything but work?”
My job is to carry a gun and fight terrorists. To fight until my energy runs out, then grab a gun again as soon as I’ve recharged, and get back on the front line. Outside the rain is pelting down, and today’s forecast is for sweltering heat all day long. I do the job I’m given, regardless of the weather. That’s what it is to be a robot.
My cooling feature’s bad, and it looks like my power will fail soon from overheating. No one else here is breaking a sweat, I’m thinking while fighting, when three guys with the same face surround me. They must all be the same model.
The middle one talks to me. “Hey, you killed a lot of guys today, too?”
“I don’t count them one by one. What’s important is they’re all enemy bots.”
“No fun at all, are ya?”
Maybe it’s because inventors like having their fun that a robot’s cognitive software has human-like personality traits scattered throughout it. This robot’s nature is overly familiar. Of course, it fights in battles as it should.
The terrorists are keeping up a strong resistance today, and the front line won’t give. My power’s on the blink.
Another of the three speaks. “Shit, nothing’s a bigger hassle than terrorists.”
“I’m sick of this. Alright, let’s shake things up. You, go lure ’em out so we can strike.”
“Hold up, why just me?” I ask.
“Say we triplets go and all wind up dead, the national media’ll crucify the government. But if it’s you, and you die, there won’t be anyone who’d mourn you.”
The Sato-Matsuzaki Lab’s process involved first selecting a model author. In choosing someone as prolific as Hoshi Shinichi (1926-1997), who wrote 1000+ short stories and essays on how to write sci-fi, they had a large source library — a necessary feature for obtaining highly variable text segments. They then wrote a rough story outline which reflected Hoshi’s general themes, and extracted text related to that outline from the library. These grammar-free text segments were entered into a sentence and paragraph generator programmed with certain parameters (e.g., sentence 1 = setting description; sentence 2 = character description; sentence 3 = action or event), producing about 300 words’ worth of grammatically correct, natural-sounding text. By repeating this step several times, sets of text were produced which were then sequentially arranged to form a recursive story that was minimally edited before being entered in the contest.
While the outline of “My Job” was developed by human researchers, its details were left up to the AI. Despite (or perhaps because of) the random nature of text segment selection and paragraph generation, the resulting voice of the piece closely resembles Hoshi’s style. Part of the program can be seen in action at http://kotoba.nuee.nagoya-u.ac.jp/sc/gw2015 (in Japanese), where anyone can freely generate a new flash-fiction piece with the press of a button.