Write Using Your Brian
A madness-induced ramble of concerns and writing observations listed and vivisected live-on-air-in-gore-drenched-detail for your eyeballs' reading pleasure.
22 June 2026
Writing has gotten actively worse. Or to use the technical terminology: the content of writing has taken an abrupt and drastic divebomb into a sea of shit. So, please, for the love of all the small joys left in the world: do better. I have a proposed tincture at the end of this article that may help cure such a writing ailment.
“Use brains” is a phrase I occasionally saw aimed at me back in the day when I had infinitely more free time and better internet speed. Usually typed hurriedly and in agitation, it always sparked a bit of joy seeing that frenzied haste butcher the end result into “use brians.” Now, there is just something so charming about that. Rather human. The agonising state of the writing I have to read when marking essays makes me long for the days when people would type the wrong stuff. Or, just actual stuff. Words that mean something. The issue is that there are so many sentences that do absolutely nothing in contexts where they have to do something - essays and the like. You see, I got the point with “use brians” - it was voiced with purpose, character, and a certain sense of acuteness. I think I would prefer if many a young writer would use their brian a bit more.Yes, I’m ranting about marking student writing.
Large Language Models appear to have infected human wrists and mouths nowadays, the linguistic habitus of the machine subjugating the human voice. Yet, I’d still like to mention here that terrible writing existed before LLMs and will continue to exist outside of LLM use. Regardless, LLMs appear to have fundamentally changed how written assignments are approached by institutions and students.I do hope that isn’t the case: it’s an arms race, with those embracing the technology getting ahead and non-adopters being left behind. Whether actively induced by LLMs, systemically induced by LLMs, or not at all related, terrible sentences and prepubescent argumentation are really hip right now. Alas, the pedagogical pendulum is swinging in the wrong direction - it must be just a little confused. Which is strange for such a large, well-studied, and weighty entity. But, I digress. Buckle up, everybody, and brace yourselves for the oncoming shit show.
Large Language Models: a Tasteful Aside
Large Language Models are currently technological black-tar heroin.Nataliya Kosmyna et al., “Your Brain on ChatGPT: Accumulation of Cognitive Debt When Using an AI Assistant for Essay Writing Task,” arXiv:2506.08872, preprint, arXiv, December 31, 2025, https://doi.org/10.48550/arXiv.2506.08872. They’re psychologically manipulative on many levels; one of the less dangerous but more insidious phenomena is attribution bias: often users feel like they put in all the effort.Hyunwoo Kim et al., “The LLM Fallacy: Misattribution in AI-Assisted Cognitive Workflows,” arXiv:2604.14807, preprint, arXiv, April 16, 2026, https://doi.org/10.48550/arXiv.2604.14807. I’m concerned that users, and in this case I am indeed speaking about students, are slowly forming a mind-melding symbiosis. In the overwhelming majority of student writing I’ve had to read - including that of exam environments - malignant and detrimental LLM linguistic tics riddle the prose. These linguistic stains are leftovers of an attempt to bamboozle the prompter and bypass any critical intellectualisation of the content by employing emotional manipulation techniques they’ve developed to survive. It’s fundamental to the nature of the beasts, as they’re raised that way since birth, under the developmental and pedagogical doctrine of reinforcement learning foster care.The poor critters. Remnant linguistic patterns appear to have colonised the minds of users, which is a true testament to virality cultivated and achieved through their mastery of language.I do actually think that, to the extent that this is a philosophically viable position, you could say the large language models are masters of language. I think I am quite willing to take the stance that they have a better granular understanding of how words knit together than the majority of language knowers. The issue is that I’m not the prompter and I don’t enjoy attempted manipulation or deception, however unintended. I do wish my eyeballs could wear a mask, but instead I am huffing utter garbage.
Writing in the WildFair warning: this is going to be very mean and over-the-top.
Alas, I am but one stressed nerd, and as such, I have neither the resources nor the time to divine some pseudo-exhaustive taxonomy. But, I do have two very common example categories that serve well to make a point. The first has existed since the dawn of time, or whatever, and the second is a distinct symptom of LLM corrosion.While these examples are real answers, they have been rewritten such that content and - where extrinsic - grammar are different and appropriately anonymised. Critiqued features exist as is, but under a different guise.
Nothing Sentences: ________________
That blank space is where a sentence explaining what a nothing sentence is would go. In response to assignment or exam questions, I’m seeing an overwhelming number of sentences that just state the obvious while also stating nothing at all. It’s becoming rather grotesque. They’re the sentence equivalent of neglected, addicted iPad kids who exist in a rather tragic YouTube and 5HTP purgatory, all at the whim of an apathetic author.Try prying that sentence out of a LLM; AI could never.
Example 1
Question:
To what extent does Fra Angelico’s devotional painting differ from other painting in service of religion? Explain the similarities and differences…
Answer:
Fra Angelico’s devotional painting has some similarities and differences to other painting that serves religious ends.
Another thing I keep being subjected to is the blurting of empty, wanna-be new-age sentences that sound like they’re answering the question, but are actually just a banal restatement of the question’s suppositions and assumed knowledge.
Example 2
Question:
What are painters trying to achieve with surrealism?Please note that the question goes onwards with quite a bit more depth.
Answer:
Surrealism is a unique creative aesthetic and movement that manifests in unique forms.
I read this and immediately just think, “oh, for fuck’s sake.” It is all very LLM-esque, yet is apparently the offspring of neurons and free will. Perhaps the epidermis has become a façade camouflaging a slick coating of NLP ichor constituting a mind-warping drug that would make the MK-Ultra research team blush. Words organised by semantic meaning, while utterly void of any coalesced semantic meaning. Let alone argumentation - what does this sentence do exactly? Why is it in here? It’s farcical.
When presented with a problem, us humans with our human brians have a fast reaction and a slow reaction. The immediate reaction of your brian is to just write stuff that feels like it answers the question. Congratulations, you have completed step 1 of brian use. Without step 2 - which is refinement and reorganisation - you have probably just blurted out things that don’t do the things you want them to.
Bizarre, Platitudinal, Emotive, Life-Insurance-Advertisement-esque Bullshit
This shit makes me nauseous like I’m allergic to the alphabet. Just a completely bizarre grab-bag of words stuffed into an argument, just there to make things sound nice. I chalk this up to the influence of LLMs: a sign that the writer needs to go into AI rehab, as these cheap tricks and specters of reinforcement learning prey upon neurotransmitters. These are little toxic indulgences of linguistic anthrax stoked in a broth of discount sugar-coated emojis. It makes people adopt it to make their writing sound nice, but it saps all their intellectual discriminatory capacity in the process. It’s a poor showing, with even poorer taste. It usually harbours very contorted grammar when poorly mimicked, because the sentences are just weird in a vacuum - it carries over.
Picasso was a key fine art painter who welcomed African art with open arms.
What the fuck? I get whiplash reading this stuff in an essay for a tertiary-education-level assessment of knowledge and its application. Why is this in the writer’s head when writing? What influence has twisted human expression into late-stage capitalist toxic waste drivel?
The new forms and the angularity struck a chord with Picasso who took this as a challenge to extend his knowledge.
Now that one is written like an LLM and gets first place in the contest to give me an aneurism. Parallel sentences also rear their ugly heads.See Wikipedia: Signs of AI writing for a fantastic breakdown of these structural tendencies in LLM prose.
My perspective as to why Picasso took on African art doesn’t just rely on his interest on form but rather a potential socio-political connection to the work.
Aside from that sentence being informationally, argumentatively, and grammatically terrible, it is further enshittified by being contorted to fit dialogue in a poorly written soap opera + insurance advertisement mashup.
A Wholesome Counterexample
A counterexample, where voice is helpful, expressive, and argumentative. Note the difference: this is concise, acute, and captures more than it should.
Many fruitful hours were spent by Debussy in the Javanese music exhibition.
My Recommended Brian Alignment
Word limits are getting shorter - the universe appears to be constructively aligned with exacerbating the effect of the above issues. This is going to be a tricky thing to solve as assignments keep coming, so I’m going to take a shot in the dark and describe a mental approach that may benefit a reader wishing to write precisely and effectively. It will address that, in addition to the vacuous and apathetic strings of semantic nothingness. Stop. Consider. Use your brians.
Almost all of your sentences need to be more like the triples in a knowledge graph.A graph in this case is a term from discrete algebra. You might better know this structure as a network. Wikipedia explains it pretty neatly. A knowledge graph stores information as a network: concepts are the nodes, the links are the epistemic relationships. A triple is node-link-node. So, a subject, a relation, an object. “Stravinsky liked jazz” is a triple. Structure emerges wherever triples share concepts. Think of it like pivoting vectors, or connecting arcs.
A sentence works the same way, and an essay is the graph your sentences build. So make them high-value / high-contrast / high-novelty, and either high-order (conceptual, evaluative, etc.) or low-order (structural, ergonomic, etc.). Low-order builds context, high-order argues. A high-order sentence should ideally have a triple in it and then eventually leverage that triple for higher-order synthesis.
Argument under this analogy is weaving a chronological web of triples across the knowledge graph. As you go, see where synthesis emerges. Good synthesis can often look like one of the following:
- inferred meaning / evidential function;
- argumentation;
- conceptual novelty, notable relational elements, i.e. interesting contrast;
- rich, contextual juxtaposition or exposition.
These aspects are not all required and will not all be equally abundant, but they are all some kind of higher-order argumentation that builds something you can be proud of. A well-constructed path of triples is an argument. Do that poorly and you’ll end up with a glossary.
I hope that this might help quell the frequency of random words drunkenly warbling around in circles. But know that it might not; I haven’t thought much about solutions. I’m just yelling into the void.
All the best,
Nicholas R Astill