Jay Kang
Jay Kang runs SEO RANK SERP, LLC and built the Anti-AI Slop Master Suite. The lexicon behind it was assembled from 21 sources, including the PubMed excess-vocabulary work published in Science Advances, the COLING 2025 paper on lexical overrepresentation, and phrase-frequency data from Pangram Labs. Every multiplier quoted on this site traces to one of them.
What this site is built on
The lexicon behind the Anti-AI Slop Master Suite was compiled from 21 sources. The quantitative anchors come from peer-reviewed corpus linguistics: the excess-vocabulary analysis of 15 million PubMed abstracts published in Science Advances, the COLING 2025 paper on why models over-select certain words, and the Stanford detector-bias study in Patterns. Phrase multipliers come from Pangram Labs.
Every figure quoted across this site traces to one of those sources, and each guide links to the paper it draws from. Where a number could not be verified, it was removed rather than softened.
Guides
AI Detectors Flagged 61.3% of Non-Native English Essays as AI
A Stanford study ran seven commercial detectors against TOEFL essays written under exam conditions. More than three in five were wrongly called machine-written. Native-speaker essays scored almost perfectly.
Read it →The AI Script Retention Cliff: Why Your Video Dies in 30 Seconds
An AI-written script fails before anyone judges the writing. It opens by restating the title and promising what is coming, which is exactly when viewers leave and the algorithm stops recommending you.
Read it →Why ChatGPT Uses So Many Em Dashes (And How to Make It Stop)
Models drop four to eight em dashes per page, which is roughly ten times the rate of edited human prose. Here is where the habit comes from, what the safe threshold is, and the instruction that actually removes it.
Read it →Google's Scaled Content Abuse Policy: What Actually Triggers It
Google named mass-produced content as spam rather than a ranking factor. The policy does not ask who wrote the page. It asks whether the page says anything, which is a harder test to pass than most publishers assume.
Read it →How to Spot AI Slop: 12 Tells, Ranked by Reliability
Not every AI tell is worth trusting. These twelve are ranked by how often they are right, starting with the two that survive editing and ending with the ones that get innocent writers accused.
Read it →Why Your LinkedIn Posts Get Flagged as AI (And What Reach You Lose)
LinkedIn added a native report control for AI slop. The formatting habits that trigger it are the same habits every engagement guide told you to adopt, which is why good posts and machine posts now look identical.
Read it →What Is AI Slop? The Measurable Definition
AI slop is machine-written text that is fluent, confident and says nothing. Here is where the word came from, the four signatures that identify it, and how to check whether your own writing has them.
Read it →Why Is AI Slop Everywhere? Four Mechanisms, Not One
AI slop is not a story about lazy writers. It is what happens when the cost of producing text falls to zero, models are trained to prefer a specific register, and platforms pay per upload.
Read it →Words ChatGPT Overuses (With the Replacements That Actually Work)
Thirty of the most over-selected words in machine writing, each with the measured multiplier and a concrete replacement. Plus why swapping words alone will not fix your drafts.
Read it →Contact
LinkedIn · [email protected] · SEO RANK SERP, LLC, 8 The Green, Suite B, Dover, DE 19901