"Coffee Break" The Phantom Reviewer and the Glass Laboratory / The Human Counterattack of Researchers Exposing the Void of AI Hallucinations

Introduction

The void known as generative AI is creeping into the sanctuary of academia: the peer review process. This story depicts the reality of an unprecedented "AI peer review hazard" faced by chemists, capturing the breath of the workplace and the comically gritty nature of humanity. It explores the malice and negligence swirling behind data and logic, and the struggle of the researchers who stand up against them. A cold yet passionate documentary novel that questions the future of science and human dignity.

The Phantom Reviewer and the Glass Laboratory

The Human Counterattack of Researchers Exposing the Void of AI Hallucinations

Body

Delusions floating on the screen in the dead of night

Two in the morning. The fluorescent lights of the laboratory hum faintly.

The fingertips tapping on the keyboard stop abruptly. The corners of the eyes staring at the screen are hot.

A single PDF file displayed on the monitor. A notification of peer review results arrived from the editor of the academic journal.

The string of characters written there makes it hard to breathe.

"Do not publish"

The harshest verdict of rejection.

My hand scrolling through the screen trembles slightly.

Comments from Reviewer 1.

As I read through the content, a cold sweat runs down my back.

It is not anger. It is confusion and an inexplicable sense of eeriness.

"Regarding the H2D+ peak in the mass spectrum of Figure S1, a blank measurement should be performed to identify the source of the peak."

My fingertips turn cold.

I look over my manuscript again and again.

What I included in Figure S1 was nuclear magnetic resonance, or NMR spectra.

There is no mass spectrum included anywhere.

In the first place, a chemical species like H2D+ should not be generated under standard mass spectrometry conditions.

Do they think I am researching the extreme conditions of outer space?

A voice echoes in my head.

A cynical laugh escapes me.

Reviewer 1's attacks do not stop.

Harsh criticism of solid-state NMR that I didn't even report.

Persistent questions regarding the fitting of X-ray fluorescence analysis that I didn't even measure.

Abuse confidently hurled at data that does not exist.

This is not text written by a human.

Hallucination.

The confident lies, the hallucinations, unique to generative AI.

The reviewer fed the paper into an AI.

And without reading the nonsensical text it spat out even once, they copy-pasted it and sent it back.

Or perhaps, they maliciously prompted it to output the 'lowest possible rating'.

I grit my teeth. My nails dig into my palms.

Data that researchers have built up with blood and sweat.

The moment it was crushed by the babble of a machine that never even read it.

In the silent laboratory, only the sound of my own ragged breathing echoed.

The Distorted Scale and the Silent Protest

The canned coffee on my desk has gone completely cold.

I re-read the evaluations from the other two reviewers.

Reviewer 2 recommends acceptance after minor revisions.

Reviewer 3 suggests reconsideration after major revisions.

Both are favorable. They appreciate the validity of the experiment.

The eyes of decent researchers.

Traces of them having sincerely engaged with the paper.

However, the editor's decision was 'reject with permission to resubmit within six months'.

A single, unjust, scathing review that abused AI.

The editor was swayed by its harsh tone.

The word "unfair" is too mild.

Is this harassment from a competing research group?

Is it a stalling tactic to delay our publication so they can release their results first?

Delusions race through my mind.

My vision blurs.

I will not take this lying down.

I place my hands on the keyboard.

At first, I tried to write a long response letter refuting every single one of Reviewer 1's incoherent points.

The keys click loudly as I type in a fit of rage.

Then, my advisor walks over.

A hand is placed firmly on my shoulder.

Silent pressure.

"Stay calm."

The professor's voice is low.

"The editor won't read a long-winded emotional outburst. Present only the facts, and make them sharp."

Deep breath.

I cool my anger and strip away the excess words.

I craft short sentences like sharpened blades.

"The criticisms from Reviewer 1, who recommended rejection, are all directed at data that does not exist in this paper. The tendency for hallucinations characteristic of generative AI is evident."

"The points raised by Reviewers 2 and 3 are experiments that can be easily addressed."

I press the send button.

My finger clicks the mouse with force.

The email vanished into the darkness of the night.

Overturned Verdict and Cold, Hard Truth

A few days later.

A reply arrived.

Words from the editor.

"I checked with Reviewer 1, and they replied that they did not use AI."

I can't help but scoff.

Of course they would lie and say they didn't use it.

However, the message continued.

"However, upon verification, I did indeed find descriptions in Reviewer 1's comments that are unrelated to this paper."

The verdict overturned.

Re-evaluation as a major revision.

Something frozen deep in my chest suddenly melts away.

The re-evaluation process.

AI Reviewer 1 had been quietly removed.

In its place, a new Reviewer 4 was added.

A few weeks later.

Reviewer 3 recommended 'publish without revisions'.

The newly added Reviewer 4 recommended "Accept after minor revisions."

The decision from the editor.

Minor revisions within two weeks, and then—acceptance.

I lean back deeply into my laboratory chair.

I look up at the fluorescent lights on the ceiling.

I won.

However, it is not just a sense of accomplishment that remains in my heart.

A deep, heavy sense of fatigue.

If I hadn't appealed directly to the editor.

If I had taken that hallucination at face value and fallen into despair.

The paper would have been buried in darkness.

A prestigious journal with over thirty years of history in the chemical community.

Behind that tradition, the peer review process is beginning to go awry.

Peer review for the advancement of science.

The folly of handing over a paper poured with passion to the ignorance of machines and the negligence of humans.

An era where human malice attacks while wearing the mask of technology.

What on earth are we fighting against?

The void of machines and the words of humans

Using AI for peer review.

Major publishers are clearly raising their voices in prohibition.

Science magazine.

Prohibits uploading submitted manuscripts to AI.

Nature magazine.

Strictly forbids uploading manuscripts and AI-generated peer reviews.

American Chemical Society (ACS).

Does not tolerate the use of AI in peer review under any circumstances.

The reasons are clear.

First, AI is incapable of the high-level logical thinking required of experts.

Second, inputting unpublished papers into AI is a fatal breach of confidentiality.

However, the field is exhausted.

A lecture by an editor at Science magazine.

The fact that even in top-tier journals, there is an increase in comments showing signs of having been written by AI.

That is why cross-review was introduced.

A system where reviewers evaluate each other's comments.

Absurd AI-generated text is exposed by the eyes of conscientious reviewers.

Anyone suspected of misusing AI even once will never be invited to review again.

Permanent banishment.

That is the retribution for researchers who abandon their integrity.

On the other hand, malice on the part of submitters is also deepening.

The rise of 'AI slop,' where papers are written by AI and submitted without even being proofread.

Editors and reviewers are screaming under the flood of low-quality papers.

Cleaning up the trash scattered by generative AI with AI-based screening tools.

The emergence of tools like Veracity and AIchemistReview.

Checking the existence of references and verifying consistency with claims.

A quagmire of an AI war.

A cold, technical slugfest between authors and journals.

Is there any passion for science left in it?

What is peer review?

It is a constructive discussion to reinforce the author's research and lead it to greater heights.

For reviewers, it is an honorable duty to encounter cutting-edge knowledge and contribute to the advancement of science.

When humans abandon thought in the name of efficiency,

what remains is a ridiculous farce danced to the tune of non-existent data.

Is there a living will in the fingertips striking the keyboard?

We are now being tested on the dignity of the narrative called science.

Seo's 'Critical Point': Cases showing signs of having AI write peer review comments have recently begun to appear.

There is a man who accepts peer review requests from top-tier journals without hesitation, yet never even opens the screen.

A printout of a paper sits on the edge of his desk, covered in dust.

The man enters a prompt.

'Critically evaluate this paper and create reasons for rejection.'

Fluent, polished English leaps onto the screen.

The man twists his mouth in satisfaction and clicks "Send" without reading it even once.

This is the true nature of the "critical point" occurring in the field.

Busy schedules, self-preservation, and twisted lust for power.

The moment these gain the "knife that doesn't stain the hands" known as generative AI, researcher ethics easily collapse.

The hallucinations spewed out by AI are a mirror reflecting the reviewer's negligence.

Criticizing non-existent spectra and fabricating data that was never measured to tear down a paper.

This is not the imperfection of AI, but irrefutable proof that humans have dumped the "responsibility of thought" onto machines.

Behind the journals that boast scientific authority, the sacred ritual of peer review is degenerating into a "throwing match of automatically generated trash."

This malicious negligence is the greatest pathology eating away at the modern academic world from the inside.

<End>

[Disclaimer and Copyright]

• Fiction: This work is fiction and has no relation to any real persons, organizations, or events.

• Copyright: Copyright belongs to the author (Yutaka Seo). Unauthorized reproduction, duplication, alteration, or commercial use is strictly prohibited. (© 2026 Yutaka Seo All rights reserved.)

• At your own risk: Please take any action based on the content of this article at your own discretion and responsibility. All views expressed here are solely those of Yutaka Seo.

• Regarding horse racing: This does not guarantee actual race results. Please enjoy horse racing at your own risk.

• *Please point out any typos, misinformation, or questions you may have regarding the article.