Posted on
by

3 oh my god

1. Wot I’ve Been Doing (permalink)

Brologue’s turning three, huh? When did that happen?

https://brologue.net/2023/11/03/prologue/

Oh. Hmm. How should I begin, person or bot who may or may not have this thread in their myriad tabs? Thank you? Hypothetical Reader, it’s hard to thank you properly when I don’t know who reads this blog. Unless you, Hypothetical Reader, take 0.000025% of your day to like, repost, and, hell, maybe even comment about these posts, even if you have little to say about them, not much is going to change:

https://brologue.net/2025/01/06/its-the-attention-economy-stupid/

I will still grind out a post every month (or three), as is my wont, so long as I can afford to keep the site running. And if my posts have entertained or, hell, even educated you, then you should definitely consider leaving a small tip in the Ko-fi jar I’m rattling right now:

https://ko-fi.com/bigbm

So, let me reflect on last year’s anniversary post and tell you what’s changed. Let me pretend there are people who are actually reading this.

First: I finished my thesis! It is, objectively, the best thing I have ever written, but treat that best as having all the energy of a six-year old who’s just seen the coolest shit ever on their iPad, and is now going to obsess about it in everything they do. I’ve still yet to grow.

My thesis is the second half of a series of tales set in the world of Boardmier, of which two tales (not part of the thesis) are available for you to read right now:

https://brologue.net/2024/12/05/the-talisman-of-aubaum/

https://brologue.net/2025/02/28/the-right-to-krei/

Don’t want to read them? That’s fine. Unfortunately for you, I’m still going to post my pitch:

Boardmier’s got familiar low fantasy trappings galore, but it’s a weird place.
For two thousand years, the Two Monarchs system has joint-ruled Boardmier, dividing its people into six castes—Pawns, Knights, Bishops, Rooks, Queens and Kings. It’s flawless—if you just ignore all the Suits who invent things and are good with money… and the impossible Undead with their unfinished business. This perversion of the natural order ought to cause outrage, but most ordinary folks yield with all the discontent of an itchy nose. It’s a slow, small world, oscillating between rigid, orderly rules and fantastique, carnivalesque rule-breaking. Whatever it wants to become, it must move forwards.

The story has really changed since these first two tales – a lot of re-tweaking is in order. The plan is to leave the story behind for a while, then pick it up again at some point next year, and work on expanding it/revising it until it’s ready to go off to a publisher. Right now, I’m waiting to read what my examiners have to say about it – their feedback won’t be ready until the first week of December. As for creative projects, I’ve got two stories ticking over that I want to tell you all about, but I shall have to vaguepost for the moment. They’ll be done when they’re done, and I will try to hawk them to any magazine1 going.

Having worked on Boardmier for so long (all of the workshops I did with my MFA cohort focused on this world), I was half-afraid that I’d struggle to shrug it from my shoulders and move on to something else, but this hasn’t been the case. Since July, I’ve been once again cognisant of just how useful the Pomodoro technique is for getting work done, especially writing, and so I’ve been using it to get started on my next projects:

https://brologue.net/2024/01/06/thought-i-d-something-more-to-say/

Can’t bring yourself to do anything creative? Give it a try, sometime.

Boardmier has been a stomping ground for figuring out what I want to be as a writer, hence the hyperactive six year-old imagery. My craft is always maturing, and greater things shall come. I hope that by the time GBH and Kreintogap are done and published (yes, that’s right, published! In real magazines!), they will stand as such greater things.

That’s pretty much all the updates I’ve got on the writing side of things. Now, what else did I talk about in my last anniversary post…?

An LLM can neither reflect nor act upon the world.

Ah.

2. Revenge of the Job Application Rejection (permalink)

Yep, it’s time to talk about AI again. Things have gotten a bit more complicated since last time, when I vibechecked the idea that “AI is plugins”:

https://brologue.net/2026/08/10/ai-is-plugins/

My opinion on it now is largely the same: take away the venture capital bubble, and the monopoly, and the data centres surplus to requirement, and I reckon we would talk about LLMs the same way that we would a software plugin.

Since ChatGPT came out, I have continuously asked myself: what can this technology do for me? What should it do? What experience do I lose by getting it to do the thing? Can I practice what I preach and justify to you what mental tasks I think are fine to shunt off to the bots? Well, after four years of answering, “nothing I couldn’t do myself,” I’ve come around on it. Yes, there is something that I would prefer a chatbot do for me.

That’s a very dicey thing for a writer to say publicly. AI is in such bad odour among writers – chemically bad odour. Whether we talk of scraping or of threats to replace us with chatbots that can (supposedly) 100x a publisher’s revenue and output, writers fucking hate AI. So, right away, I want to reiterate: getting these things to write my fiction/blog posts for me is out of the question. If I’ve got something to tell you, you’re going to hear it from me. That’s the point of writing, despite the weird, unsubtle, volatile, divisive, dogshit, diabolical, incredibly dialectical times we live in. The metaphorical House of Context may stand on chalk pillars in a lake of vinegar, but I will still try, despite my flaws.

Just because these are principles I hold and am telling you I believe in, that doesn’t mean I’m any less immune to the allure of using AI technologies to avoid interacting with other people. It’s like someone telling you they’re not racist. Trying to prove that someone holds racist beliefs mid-conversation is a fool’s errand, but you can reasonably prove the things that someone says that are, whether they believe it or not, racist. So, even when I tell you my principles, you, Hypothetical Reader, have a right to show me where I say things that might unintentionally align me with people who would use AI to make the world a worse place. Not just weird chuds on Twitter, but the doubly weird, rational, TECSREAL types on Bluesky, who plug chatbots into a handle, and would really like us to believe that these things are people, not a file on someone else’s computer. I can at least laugh at the chuds.

I think it’s fair to say that the things about AI that you and I hate are downstream of capitalism. Capital did not produce AI, to paraphrase a recent Brendan Keogh post – human knowledge and labour did:

https://brkeogh.com/2026/11/11/ai-is-history

To adapt a thought experiment from the late David Graeber: suppose we all woke up tomorrow, and every one of us decided to stop participating in capitalism. Every technology that has been branded as “AI” over the past eighty-odd years would still remain. We’d have to figure out what to do with it. While it is cathartic to imagine a future where this technology just goes away entirely, I am persuaded that an all-purging Butlerian Jihad shall not come to pass in the material world in such a way.

We do not defeat capitalism by “turning back the clock”, by a life of abstinence and “artificial self-limitation” but by moving beyond capitalism and fully developing its “civilising tendency” beyond a mere byproduct of the profit motive into the actual motive force of our collective society focused on the collective good.

I think any robust critique of generative AI use in game development (or indeed any field) needs to be situated within history and not outside of history. It needs to recognise when its critiques of generative AI are actually just deflected critiques of capitalist relations of technology and labour […] what is a criticism of the ‘technology itself’ and how this technology is used and shaped by capital to exploit labour and increase profit at any cost […] Once this distinction is clarified, then a conversation about the potential uses of generative AI (if indeed there are any) can be had.

Brendan Keogh

In my mind, I thought I was patient enough to wait until the bubble bursts, that then it would be “safe” to talk about what gen. AI “can” and “should” do for us. In the material world, however… Well, no use prevaricating around the bush. Are we agreed – I haven’t succumbed to AI psychosis? I hope so. Now, what can this supposed machine god, confined to a 17 GB file on my computer, do for me today?

Write my CV.

I have a caustic, spiteful hostility towards that thing we call the job’hunt.’ We make it sound so twee. The last time I checked, jobs aren’t birds for me to shoot up for Christmas dinner, in my tweed and deerstalker; they aren’t vixen, to be hounded to death. People say you shouldn’t take rejections personally. Oh, I don’t. I see them as developments in the perpetual war of attrition I’ve been needlessly drafted into. Hypothetical Reader, I have fannied about in the dark long enough with this shit, panning the river bed for gold-nugget keywords, and so have you. I get tremors in my hands from this shit, man, every time I’m rejected. How many precious hours of your life have you lost fumbling applications because, “UH UH UH! YOU DIDN’T SAY THE MAGIC WORD!”? How many hours, despite the fact that you said all the magic words?

When we touch up our CV with everything we’re told we should be doing, we believe we’re arming ourselves with more lucky charms to ward off NOREPLY. And the things recommended to us are lucky charms – you bless ’em when they work, and curse ’em when they don’t.

https://brologue.net/2024/04/25/stare-at-screen/

Anything we do that we believe will make our CVs look more attractive… it’s all cargo cult shit to me. We are trying to get the attention of an alien colony organism’s immune system, begging to become part of the gut flora within its colon. If one of these things notices us, I’m convinced it can only be chalked up as a sheer freak accident of luck.

There’s an infinitely citeable quote about LLM-extruded text, that often makes the rounds on social media. It might have originated from Prof. Emily M. Bender, linguistics professor at the University of Washingon, I’m not sure – you might’ve heard it from a bunch of different people – but she, more than anyone else, is the person who I want to cite as having said it. Her name carries a lot of weight and will, I hope, produce a certain filtering effect on the audience of this post:

Why should I bother to read something you couldn’t be bothered to write?

It’s a brilliant quote, and it ought to count for the greater 90% of everything that is shit just as much as it should count for the golden 10%. The dark corollary to it, however, is this: Why should I bother to write something that a pair of human eyeballs hasn’t been asked/couldn’t be arsed to read first? The past is a foreign country and they do things differently there. Like getting humans, and only humans, to read CVs and make a judgment call. I am envious of you, job-havers, who broke your toes in the door-slammed thresholds of industry the old-fashioned way, the pre-2022 way. Now, it’s the future, where your potential boss has got ChatGPT on speed dial, and it tells him that it loooooooooooves reviewing CVs like flies landing on shit. And you know what? Good. Let my potential employers use AI for screening, on the condition that we get to use LLMs to write our applications. Fair’s fair.

Now, when I say I want a chatbot to write my CV, I don’t mean just asking it confabulate That Time I Definitely Got Hit By a Bus and Was Isekai’d to the Timeline Where I Was NASA’s Youngest Software Engineer, Aged Three2 (And Made It Back Again). I just mean that I’m a person with a diverse lot of interests/skills to spin, for two pages of CV. I’ve done a lot of things I would count as “noteworthy achievements.” I’ve had one foot in tech and the other in the humanities – it’s fair to say that I am an unusual, protean individual who would be a great asset to many an employer. I am also AuDHD, and, try as I might, even with the advice of job coaches, my CVs don’t make it past screening. I do not need to know how an employer’s applicant tracking system works to know this. The purpose of a system is what it does.

Hypothetical Reader, believe me, I have fucking tried, and tried harder.

I do believe that in any domains in which you are an expert, you can get an LLM to cough up reasonably useful results. Or, at least, you as the expert have greater discernment over an LLM’s outputs. And as the expert on everything I’ve ever done3, and the expert of my past work experience, I’ve been more than happy to let a bot write my CV for me. I have better things to do with my time on earth.

I’m not the only one who feels this way. The BBC recently ran an article about graduate students who use AI to write their CVs:

https://www.bbc.co.uk/news/articles/c623dnkn8pkyo

It’s claimed that 29% of employers plan to introduce AI into their recruitment processes next year:

https://ise.org.uk/knowledge/insights/615/graduate_jobs_stabilise_but_ai_use_disrupts_recruitment

Really. Let’s assume that’s the best case scenario. Does this mean I should unironically assume a greater 71% of goody-goody employers aren’t using AI? Pull the other one, it’s got bells on. In my book, 29% = 100%. If one company uses AI, you can bet they all do; if they use AI in their recruitment process, then I’ll use AI right back.

Is this, as business undergrad Bhuvana Chilukuri put it, lazy?

https://www.bbc.co.uk/news/articles/clyznzrqeggo

If she meant asking a chatbot for a CV, sprung from the head of Zeus fully formed, then yes. It’s far less lazy, however, and more truthful of the skills you actually possess, if you develop and manage a compost heap full of source material files. That’s what I’ve tried to do, based on the advice of Alistair Davidson, who I would vouch has a clearer head on using AI for progressive ends than your average poster:

https://bsky.app/profile/brologue.net/post/3mwvvj7timk25

In my Great Automatic CV-inator’s project files, I’ve got a folder containing ten Markdown files containing info about these key areas (in no particular order):

  1. Contacts: Every method of contact an employer would need;
  2. Professional summary: Very brief introduction about myself, businesslike tone;
  3. Previous occupations: Self-explanatory;
  4. Education: Secondary school, my ethical hacking undergrad, my creative writing MFA, and lists of noteworthy things I did (e.g., the fantasy workshop I ran for the undergrads);
  5. Personal achievements: Any recognised personal achievements outside of work;
  6. Soft skills;
  7. Technical skills – all things software, really;
  8. Publications: Brologue posts; things that show I’m well-read and well-written;
  9. Certifications: Any certifications I’ve achieved through training;
  10. Hobbies and interests.

I asked the agent I’ve got running in the OpenCode harness to write a command that, when called, will curate this folder for substrings that best match the job description provided. These must be verbatim, from the source material files, not fabricated. You might reasonably expect LLMs to make a complete dog’s dinner of this, as I did at first. All I can say is, of the CVs I’ve made with it so far, not a single one has fabricated something that’s made it into the final PDF.

Of course, when it comes to hitting keywords, there needs to be some room for paraphrasing, and this is something an LLM can also reasonably do. It’s not a jumped-up autocomplete for nothing, you know? That’s why, later in development, I asked it to develop a spin doctor subagent to identify areas that could be better paraphrased to match the advertisement. Sometimes this leaves some dangling sentence fragments, and so, per its instructions, it asks me to approve any changes before making them.

Whether all this effort will land me an interview or not remains to be seen. This could all just be yet more cargo cult type shit, I don’t know. A paper last year states that callbacks based on cover letters have fallen by 51% around about the release of ChatGPT-4 (at least, that’s how I understand it):

https://arxiv.org/html/2509.25054v1

In response to AI-tailored CVs, presumably employers have already begun to prioritise these other “signals,” like previous employment, reviews, and so on. It’s a war of attrition, like I say. You poke me with a stick, and I’ll poke back and prepare for you coming back with a bigger stick.

I’m aware that, without anything concrete to show you, all of the above kind of comes off as insane rambling. I can’t, in good conscience, give you the prompts I wrote for this, or the additional files this project has produced over time. I just don’t think it’s appropriate for me to distribute software that has been developed for myself, for my bespoke purpose, to scratch my particular itch. That’s the trouble with vibecoding, as Ian Bogost wrote in The Atlantic back in October:

https://archive.is/MzCSO

Up until very recently, the terms ‘app’ and ‘software’ used to mean things that were, by and large, produced by collective action only. Hypothetical Reader, if you’re a coder, then tell me this: when was the last time you consulted Stack Overflow for a coding problem? Last year? Amateur or professional, I think it’s hard to deny that vibecoding has alienated us from that collective aspect. It’s great if you’re an amateur who has a very specific itch of a problem that you need to scratch – say, a small Lua plugin for your e-reader. On bigger, more complex projects, however, how can you be the expert when the LLM is solving all the problems for you?

AI has shifted from facilitating programming – as a process of iteratively MAKING STUFF – to solving a problem on its own.

If you’re distributing that code for others to use, you owe it to your users, A) to declare AI usage and specify exactly what you’ve used it for, and B) to check the code for bugs. If you are an amateur (like I am) and can’t be bothered to audit all your vibecoded slop, then putting it online is kind of like that one image of a pig shitting on its own balls. “Well? What do you mean you’re not going to clean the shit off my balls?”

What has been built for me treats a particularly furious rash that I would otherwise try to treat with piranha solution, first:

https://www.youtube.com/shorts/tGCZts3z4QQ

I am both an expert in what this thing does for me, AND not an expert enough to stick it on Github and say, “Come fill your boots!”

All I can really do now is add more stuff to my source material files. I’m currently volunteering as a retail assistant in a charity shop, so that’s going in. I will provisionally accept the advice online that volunteering helps boost the odds of my CV making it to an interview, until it doesn’t:

https://blog.scotland.shelter.org.uk/volunteering-pathway-employment/

https://jobhelp.campaign.gov.uk/improve-your-chances-of-getting-a-job/cv-job-applications-interviews/make-your-cv-stand-out/no-work-experience-focus-on-what-you-do-have

Thus, the war goes on. Brologue will go on. I will go on. Hypothetical Reader, I shall go on until you aren’t so Hypothetical anymore, and then go on some more. Writing for you will always matter to me.

  1. Although I’m particularly gunning for Clarkesworld, because every time I pick up an issue, I see just what it takes to be a published writer, and I know, with a little elbow grease, I’ll make it there. ↩︎
  2. Three years old; thirty-five years’ experience. ↩︎
  3. To clarify what I mean by “expert”: During my undergrad, I got a D in a module on Android development because I couldn’t write Java for shit. I know what I don’t know. ↩︎

TTLY… (permalink)

CC

“The Brologue.net Triple Triennial Special” is licensed under a Creative Commons Attribution 4.0 International license. So long as you clearly link back to this webpage and credit me, bm, you are free to use this post however you like. Quotations, images and screenshots not included; proceed with caution.