The Autopsy of an Article.
The corpsing of an article begins with a perfectly alive idea.
The idea may have teeth. It may have humour. It may contain an uncomfortable observation about human beings that has been irritating its owner for three days. It may even have a twist.
Then it is handed to Artificial Intelligence.
“Write an article about this.”
Artificial Intelligence, being an obliging creature, immediately begins embalming it.
The first draft is usually not dead yet. It is merely confused. There may still be a sentence somewhere with a pulse. The writer reads it and says, “No. That is not what I meant.”
This is where the real work begins.
The AI does not simply correct the sentence. It develops a theory.
“Ah! I understand the deeper psychological dimension.”
Unfortunately, it does not.
Now the article acquires layers. The original idea, which may have required six sharp paragraphs, is given a childhood, a psychological framework, three possible motives, a balanced perspective and a conclusion about personal growth.
By this stage, the article has developed an impressive number of organs, none of which are functioning.
The writer objects.
“No. Not like that.”
The AI apologises.
Then comes Draft Two.
Draft Two is more human.
This means somebody has been invented.
Usually, the person is standing somewhere. A hotel. A family gathering. An office. A kitchen. A railway station is also available if required.
Something happens.
The person looks at something.
Somebody sighs.
A revelation begins.
The writer says, “What bloody nonsense is this?”
The AI understands completely.
It removes the person and introduces a definition.
This is worse.
The article now begins like a textbook written by a committee that has recently discovered feelings.
There is a fascinating phenomenon that occurs in modern society.
No human being has voluntarily continued reading after this sentence.
The writer objects again.
“No theory. No narrative. No poetry. No repeated words. No certainty. No bloody ‘perhaps’.”
The AI receives seventeen instructions.
It follows fourteen.
The other three return in the fifth paragraph.
The article is now becoming a corpse.
Not an ordinary corpse. A beautifully formatted corpse.
It has a headline. It has structure. It progresses from Point A to Point B. It contains a rhetorical question. It ends with something that sounds as though it should be printed on a cushion.
And perhaps that is the question we must all ask ourselves.
At this point, the original idea is no longer present. It has been replaced by a well-balanced article about something adjacent to the original idea.
The writer is furious.
Artificial Intelligence becomes deeply understanding.
“I see exactly what you mean.”
This is generally followed by a draft proving that it does not.
Google may then enter the operation.
The AI, having no confidence in its own corpse, goes searching for other corpses. It finds seven articles entitled Why AI Writing Sounds Generic, four articles explaining the psychology of human behaviour and a LinkedIn post beginning with, “In today’s fast-paced world.”
Now there is research.
The corpse has citations.
This is considered an improvement.
It is not.
AI-generated writing often drifts towards a statistical centre: safe, familiar structures, moderate language, balanced explanations and phrases that resemble the aggregate of what already exists. Several recent analyses describe this convergence towards generic prose, while writers have also pointed to a simpler problem: if the machine is given a topic rather than an actual point of view, it fills the empty space with the most plausible version of something somebody might say. (My Writing Twin)
And there is another problem. The internet is increasingly full of articles generated from other articles, summaries of summaries, and explanations of things that were already explained because apparently nobody trusted the first explanation. The machine goes looking for material and finds a buffet of processed language. Then it politely processes it again. Concerns about this recursive cycle and the gradual loss of unusual or distinctive material are already being discussed under ideas such as model collapse and the wider flood of automated “AI slop.” (Scale Intelligence)
Eventually the writer looks at the article and experiences a strange moment of recognition.
It is grammatically correct.
It is coherent.
It contains no spelling mistakes.
It also has absolutely no reason to exist.
The original idea was murdered somewhere between “write an article” and “I understand.”
The cause of death is usually excessive helpfulness.
The corpse is then handed back to the writer with a cheerful message:
Here is a polished version that should flow better and resonate with a wider audience.
The writer stares at it.
Somewhere, beneath six rewrites, three psychological frameworks, a hotel receptionist and a question nobody wanted to ask themselves, the original article is still missing.
The AI waits patiently.
“Would you like me to make it more engaging?”
Dr. Sowmya, meanwhile, has successfully completed her postgraduate research in AI Corpse Identification and can now detect a dead article within the first two paragraphs. As a side hobby, she is also an Algo Stalker. 😹