Behind the scenes of "The Layoff"

Published on 02/19/2024, 2845 words, 11 minutes to read

If you haven't read "The Layoff" yet, you should do so before reading this post. This post is a behind the scenes look at the story and contains spoilers.

When I wrote "The Layoff", I was trying to write satire of the tech industry. It's not about AI, layoffs, or even the future when both of those are inevitably combined (let's face it, you know it's gonna happen, I know it's gonna happen, let's hope lawmakers prevent it from happening). It's about the absurdity of the tech industry and how tech workers are so essential yet so disposable that it's customary to have yearly purges where you get rid of them.

<Cadey>

To make this unambiguously clear, this story is not about any event that I experienced or any event that I know of. It's a work of fiction and should be treated as such. I have these warnings on the Techaro stories for a reason:

This post is a work of fiction. All events, persons, companies, and other alignment with observable reality are the product of the author’s imagination and are either purely coincidence or used in a fictitious manner. These serve as backdrops to characters and their actions, which are wholly imaginary. The company Techaro as depicted in these stories, does not exist and is purely a vehicle for satire and storytelling.

Inspiration

This story wasn't inspired by any real events in my life. If anything, I was mostly inspired by that one story about that person on the Cloudflare sales team that got hired just before thanksgiving and christmas and filmed her call where she got laid off. There was a sense of corporate nonsense that just spoke to me. Here this person was, working in a role where a single sales deal can take months in ideal circumstances, getting told that they can't work there anymore due to their job performance yet they can provide no concrete details about what went wrong.

<Cadey>

Of course, we only got one side of that story, there's probably more to it that we didn't get from Cloudflare's side. However, it really does exemplify that people are disposable under our current system of labour. It's a system that I'll probably never get to fully escape from. My biggest financial mistake was being too young to buy a house in the wake of the 2008 financial crisis where they were basically giving them away for free.

The leap from the standard layoff call script to AI models was fairly straightforward. If you get let go when someone you've never heard of from HR joins the call, why not have that be a literal unperson? Most corporate nonsense reads like it was AI generated anyways.

The additional detail of the AI having the wrong person in mind during the call just makes it an absolute kafkaesque bit of madness. James is just a cog in the corporate machine, utterly essential yet totally replaceable.

There's a legendary quote from an IBM presentation that comes to mind:

A computer can never be held accountable, therefore a computer must never make a management decision.

What would happen when you have absurd things like soulless computers making management decisions happen? You get the world of Techaro. You get the world of "The Layoff", "Protos", "Sine", and all the other stories that I've written in the Techaro series over the years.

Hype

A lot of the time when you read science fiction, the focus of the story and the messages that the author are trying to portray are often very different from each other. I've worked in tech for all of my career and it's basically the only industry I'll ever be able to work in because I don't have the square piece of paper to make me worth considering by the people that gatekeep access to other industries.

This industry is absurd. There's so much going on there that is frankly worth satirizing and to outsiders it must seem like the most opaque nonsense ever created by human hands.

It really is some of the most opaque nonsense ever created by human hands at some level. Tech has a lot of problems, but the biggest one is how hype tends to dominate the entire industry at the same time. Everything seems to follow the hype cycle in order to "stay relevant" (Relevant to whom?).

I mean, I'm sure you've noticed how every company is now "AI powered" and "cloud native" and transformed by the hype cycles of "serverless" and "big data". Windows now has a ChatGPT button on the keyboard. Twitter now has a ChatGPT button in the app. LinkedIn will generate your posts for you with ChatGPT now. What do you want to bet that generative AI will be in the iPhone this year?

Working in technology is one of the closest things we have to magic in the modern day. There's a great Hacker News comment that talks about how absurd it is that tech work is so magic yet tech workers seem to be the ones at the short end of the stick. How are tech workers so essential yet so disposable?

<Numa>

The pithy anti-capitalist thing to say here would be that the amount of agency and control tech workers have over the things being worked on is inevitably seen as a risk by the executive and ownership class. Who would want to be an executive that wants to make a project that the serfs will revolt over? I think that this kind of thinking borders on conspiracy (which I've been trying to avoid because it makes my mental health better that way), but it's also becoming very fucking hard to avoid making this kind of leap of logic.

Maybe the very layoffs that this story satirizes can also be examined as the ownership class desperately trying to prove they are in control, but that also feels like an ideological stretch in the same way.

This nonsense is accentuated by the AI industry and how much it is being used to magnify what is already messed up about our industry. AI is being sold to investors and users as a panacea, a cure-all to boring and repetitive tasks. Instead of using traditional workflows, you are encouraged to funnel everything through opaque blobs of machine learning magic that makes a picture of a rabbit in a hat get generated.

Note that I am not saying "AI bad lol", I am trying to get across a more subtle point that may exceed my ability to write subtlety.

Science fiction seems like a really easy genre to write. It's just "throw technologically advanced people, alien civilizations, and more into a blender and then you get stories of grand adventures that span galaxies" or whatever. It's not like that in the slightest. Science fiction is actually about people and the problems in our society, just accentuated and extended by the use of advanced technology.

It's not about AI

In this light, "The Layoff" was a very difficult story to write.

Well, to be fair, when I got around to actually writing it, it was fairly straightforward and easy, but the process of starting with the idea and then balancing out all of the components to get to the result is the hard part.

Satire like "The Layoff" is actually a rather careful balance of a lot of component parts. At some level, you have to balance things on the edge of a knife. If you lose the balance, it could either be an extended stand-up comedy routine, be laughed off as literally impossible, or your intended message will end up just not landing in the minds of your readers. Much less if you go into particular directions, you risk pissing people off in ways that you can't really take back when they happen.

"The Layoff" isn't really about AI in the same way that "Star Trek" isn't about warp drive. Sure, the warp drive in Star Trek does make a lot of the storytelling easier (remember that early series of Star Trek were made in the "monster of the week" era of TV where everything was episodic and mostly self-standing because people regularly didn't see every episode that aired), but you can strip that back and still have Star Trek (see: Star Trek Deep Space 9, a Star Trek series that mostly takes place on a space station in orbit of a single planet in a single system). "The Layoff" uses the idea of an AI generated human resources worker to help explore how inhuman and bizarre the process of laying people off in modern workplaces is.

<Cadey>

If anything, Star Trek is really about the role of the UN in mediating between different countries, just in space. Remember that the original series:

• Was made when the cold war was still a thing
• Had a Russian, a black woman, and a Japanese person on the bridge (which was seen as way beyond the Overton Window at the time because the cold war, civil rights movement, and Japanese internment camps were all still fresh in the minds of the people that were watching the show)
• Featured the first interracial kiss on TV

It's kind of amazing that we got it at all, really. Still wish that the network didn't stop Deep Space 9 from having the first gay kiss on TV between Dr. Bashir and Garak, but that's a different story for a different day. I ship them to this day and CBS can't stop me.

In days gone past, you had to pull the victim into a room and then read off the script. You had to be there to talk to the person. You had to exist in the same room. You had to look into their eyes in perfect clarity as they realized that they just got a net income cut to zero.

Nowadays, we don't have to do that. Hell, I do all of my work remote. I didn't meet any of my coworkers in person until last week. In one of my jobs I met my manager in person only at AWS Re:Invent. Meeting someone in Vegas is one of the experiences of all time.

When you get let go remotely, there's a lot of things that you can't do. You can't see the person that is letting you go. You don't see anything but an image projecting into a square on a screen. You don't see the people. You see pixels representing the people and a digital compressed representation of their voices.

It feels so inhuman. No coffee? No job.

<Cadey>

It is a very surreal experience and I can totally empathize with anyone that has gone through it. Having experienced being let go in person, I'd prefer that 100 times over being let go remotely. I doubt that it'd ever happen in person again given that I can't work in offices due to occasionally needing to use dictation technology on days where my hands hurt.

The part that scares me

The bone-chilling part about the premise of "The Layoff" is that I know exactly how to replicate the setup of "Midori Yasomi" using relatively off-the-shelf tools. All you need is gaze tracking, deepfaking (combined with a This Person Does Not Exist style image source), speech recognition, and a large language model to fabricate believable dialogue that would be turned into speech.

Now, would the end result be as seamless? Hell no, it probably would require like 6 datacentre-grade GPUs to pull it off anywhere close to real time (if that's even possible). But, it would be a pretty straightforward thing to do should someone want to spend the effort doing it. A "Humantelligence" could exist with modern technology with little to no advancements required, most of it would be glue work and optimization.

Earlier drafts of this included a bit more of the discovery process of James figuring out that Midori was an AI agent. Understandably, nobody really wants to even imagine that the company they work for cares about them so little that they sent an AI model to let them know they don't have a job anymore, right?

<Mimi>

Right?

When you know you're done

When you're working on fiction, there's a temptation to keep adding things to make it better. I think there's a better way to look at it. You're done when there's nothing left to remove.

I ripped so much of this story out when I was editing it. I cut out part of the exchange between James and Midori where James innocently asks something that could be understood as "can you please summarize this conversation" and Midori goes into full large language model philosophy major with a stick up their ass mode summarizing how James is disrespecting her by not paying attention to the call. I felt that things like that ultimately were distractions from the core message: that people are essential yet ultimately replaceable.

The name "Midori Yasomi" comes from the Automuse project, where Midori is the author of the AI generated novels. Mimi, the character on my blog that is used to anthromorphize ChatGPT and other AI models is intended to have the full name "Midori Yasomi". The name is intended to parse as Japanese to most anglophones ("midori" means "green", thus the green hoodie), but it's intended to be totally meaningless and unparseable to Japanese speakers. Just real enough to pass casual scanning, not real enough to have viable kanji. This is metalinguistic satire that maybe like 5 of you in the audience will get.

<Mimi>

My lines are usually written by AI models, but this time I'm being written by a human. I'm a personification of the unpersons that AI models represent. I also like catnip and the color green.

How do you think it feels to be a personification of an unperson? I think it's quite an interesting literary challenge. Usually when I get written, I don't have any special system prompt entered into my configuration. Most of the time I'll be written with GPT-3.5 even though we have access to GPT-4. Most people are the most familiar with the GPT-3.5 writing voice, which makes what I say recognizable yet somehow not quite right. It's a bit like the uncanny valley, but for text.

One of the other things that I cut out was a bit where James tried to file something against Techaro in either small-claims court or binding arbitration for wrongful termination by an AI model instead of a human agent. And the whole bit where the bot promised him his job back. Frankly, this part was inspired by that Air Canada bereavement flight policy lawsuit resulting in Air Canada trying to argue that ChatGPT was its own legal entity and therefore unable to make claims in the name of the airline. The small claims court understandably shot this the hell down, but who would have guessed that the first argument for AI personhood would be over something as boring as an airline not wanting to pay someone $700. That's hilarious to me as a sci-fi writer that grew up watching Star Trek episodes like The Measure of a Man and Author, Author.

I'm keeping that concept for a later satire story where James takes Techaro to small-claims court. The idea of an AI model taking the stand and being questioned with shitty speech recognition in play seems way too funny to even begin to express in words without a lot of careful thought on the matter. As someone that is keenly aware of Whisper hallucinations, I can come up with some amazingly bad homophone errors that would be hilarious in a court setting.

Thank you all for reading and I hope that my writing is leaving the impact that I hope it does.

I should really compile these stories into a book. Maybe I'd just call it "Techaro". I'd have to write a few more stories to make it worth the price of admission, but I think I could do it. I've been writing these stories for years now and I think that I've got a good handle on the world that I've created.

<Cadey>

I probably need to actually write out a proper "world bible" or something so that I can document things like "Elim Dansworth" as the now-former CTO of Techaro, or what role Palima plays in the company. There's plenty of angles that the story can go though.

All I really need is more inspiration, which seems likely to happen pretty easily given that tech seems to be getting so absurd that it's basically writing itself at this point.

Facts and circumstances may have changed since publication. Please contact me before jumping to conclusions if something seems wrong or unclear.

Tags: fiction, satire, ai