The Slop Taxonomy
Not all AI slop is the same. We classify it into five recognizable patterns — each with distinct signals, common phrases, and side-by-side slop vs. not-slop comparisons.
Quick Reference
| # | Type | Telltale Phrase |
|---|---|---|
| 1 | Generic Slop | In today's fast-paced world... |
| 2 | Pseudo-Insight Slop | The key is to find balance... |
| 3 | Fake Authority Slop | Studies have shown... |
| 4 | Wikipedia Rehash | X is defined as... |
| 5 | Wellness Slop | Self-care isn't selfish... |
Why Classify Slop?
🔍
Faster Detection
Once you learn the five patterns, you can spot slop in seconds — before you waste time reading it.
✍️
Better Writing
Knowing the patterns helps you avoid them in your own writing — especially when using AI as a drafting tool.
🧪
Academic Framework
Researchers use taxonomies to measure AI slop across information utility, quality, and style dimensions.
Generic Slop
The verbal equivalent of stock photos
Most commonTemplated, vague, no concrete details. This is the most common type of AI slop — content that could have been written about literally anything because it says nothing specific about anything. Researchers call this 'cohesion without coherence' — the text is grammatically connected but semantically empty.
Detection Signals
- ⚠Opens with broad generalizations ('In today's fast-paced world...')
- ⚠Uses filler phrases that add zero information
- ⚠Could apply to almost any topic — swap the subject and it still works
- ⚠No specific names, dates, numbers, or examples
- ⚠Relies on 'best practices' and 'key takeaways' without saying what they are
- ⚠Monotonous sentence rhythm — uniform paragraph lengths, no variation in cadence
Common Phrases
"In today's fast-paced world, it's more important than ever to stay ahead of the curve. Whether you're a beginner or an expert, understanding the fundamentals is key to success. By leveraging best practices and innovative solutions, you can take your results to the next level and achieve your goals."
"Between January and March 2024, our team at Acme Corp tested three different onboarding flows with 2,400 users. Version C — which replaced the 7-step wizard with a single-page form — increased completion rates from 34% to 71%. Here's exactly what we changed and why."
Why the good example works
Specific company, specific timeframe, specific numbers, specific method, specific result. Every sentence adds new information.
Commonly found in: Blog posts, LinkedIn articles, content farms, product descriptions, About Us pages
Pseudo-Insight Slop
Philosophy for people who don't read philosophy
Very commonContent that sounds profound but delivers zero information gain. Every sentence seems like it should be meaningful, but when you examine it closely, nothing was actually communicated. The literary equivalent of a fortune cookie. Charlie Guo calls this 'unearned profundity' — a distinctive mark of AI-generated text.
Detection Signals
- ⚠Profound-sounding but content-free statements
- ⚠Obvious observations dressed up as wisdom
- ⚠Vague metaphors that don't actually clarify anything
- ⚠Tautologies presented as insights ('Success comes from being successful')
- ⚠Inspirational tone with no actionable substance
- ⚠Dramatic single-sentence reveals: 'Something shifted.' 'But now?'
Common Phrases
"The journey of a thousand miles begins with a single step. Remember, it's not about the destination — it's about the journey. When you change the way you look at things, the things you look at change. True growth comes from within, and the only limit is the one you set for yourself."
"I failed my first 3 startups before one worked. The specific mistake that killed startup #2: I spent 8 months building features nobody asked for instead of talking to the 12 beta users who'd already signed up. Here's what I'd do differently."
Why the good example works
Personal stake, specific failure, concrete number of startups and users, and a defined actionable lesson — not a platitude.
Commonly found in: Self-help content, motivational posts, corporate keynotes, LinkedIn influencer posts, graduation speeches
Wikipedia Rehash
The first paragraph of Wikipedia, rephrased
Common (esp. educational content)Content that presents basic, commonly available definitions as if they were original insight. You could get the same information — often better organized — from the first paragraph of a Wikipedia article. No analysis, no original perspective, just rephrased common knowledge. Google's Quality Rater Guidelines specifically flag this as 'low-effort paraphrased content.'
Detection Signals
- ⚠"X is defined as..."
- ⚠"X refers to the process of..."
- ⚠Basic definitions presented as content
- ⚠No analysis beyond what's commonly known
- ⚠Educational tone without actual educational value
- ⚠Could be replaced by a Wikipedia link
Common Phrases
"Machine learning is a subset of artificial intelligence that enables computers to learn from data without being explicitly programmed. It has many applications in various fields including healthcare, finance, and transportation. There are three main types of machine learning: supervised learning, unsupervised learning, and reinforcement learning."
"Most ML tutorials explain the three types of machine learning. What they don't tell you: in production at our company, 90% of our models are supervised learning with logistic regression — not the fancy deep learning stuff. Here's why simple models win in practice, and when they don't."
Why the good example works
Acknowledges the basic knowledge exists, then adds original production experience that Wikipedia can't provide.
Commonly found in: Educational blogs, 'What is X?' articles, introductory guides, product explanations, AI-generated course content
Wellness Slop
The content equivalent of a live, laugh, love sign
Common (esp. social media)Safe, soothing, universalized self-help that applies to everyone and therefore helps no one. Every sentence is designed to be comforting rather than useful. No specific techniques, no challenging ideas, no personal experience — just a warm blanket of nothing. The key test: if advice could apply to literally every person alive, it's wellness slop.
Detection Signals
- ⚠Gentle, non-confrontational advice that can't be wrong
- ⚠Universalized statements that apply to literally everyone
- ⚠"Remember to be kind to yourself..."
- ⚠Wellness buzzwords without actionable guidance
- ⚠Assumes the reader needs emotional comfort, not information
- ⚠Therapy-speak without therapeutic content: 'boundaries', 'journey', 'healing'
Common Phrases
"Remember that self-care isn't selfish — it's necessary. Take time to breathe, practice gratitude, and honor your journey. You are exactly where you need to be. Embrace your authentic self and set healthy boundaries. The universe has a plan for your healing."
"After my burnout in 2023, I tried 'self-care' for 6 months — baths, journaling, yoga. None of it worked. What actually helped: I quit my Thursday night commitment, started saying no to weekend plans, and went to bed at 9:30pm. Boring, specific, effective."
Why the good example works
Personal timeline, specific failed attempts, specific successful changes, and the honesty that 'self-care' platitudes didn't work.
Commonly found in: Instagram captions, wellness blogs, therapy-speak content, motivational accounts, corporate wellness programs
How Types Overlap
Real-world slop rarely fits neatly into one category. Most pieces combine multiple types — which is why our detector reports all patterns found, not just the dominant one.
Generic + Pseudo-Insight (most common combo)
Opens with vague generalizations, then fills with profound-sounding but empty observations.
"In today's fast-paced world, the key is to find balance. True growth comes from within..."
Fake Authority + Wikipedia Rehash
Presents basic definitions with an authoritative voice, citing nonexistent studies.
"Studies show that machine learning, which is a subset of AI, is increasingly important in healthcare..."
Wellness + Pseudo-Insight
Soothing self-help tone wrapped in faux-philosophical observations.
"Remember, the journey of healing begins with self-acceptance. When you change the way you look at your boundaries..."
The Antidote: What Makes Writing Not Slop
Across all five types, the antidote is the same set of qualities. Content that has any of these is almost certainly not slop:
Specificity
Named people, named places, specific dates, exact numbers. If you can fact-check it, it's probably not slop.
Lived Experience
Personal stakes, specific failures, lessons that could only come from doing the thing — not from summarizing it.
Cited Sources
Named studies, linked references, attributed quotes. Every factual claim should be traceable to its origin.
Information Gain
After reading it, you know something you didn't before. The simplest test: can you summarize what was new?
Frequently Asked Questions
What are the 5 types of AI slop?
(1) Generic Slop — templated, vague content with no specifics; (2) Pseudo-Insight Slop — sounds profound but communicates nothing; (3) Fake Authority Slop — confident tone with no citations or evidence; (4) Wikipedia Rehash — rephrased common knowledge with no analysis; (5) Wellness Slop — universalized self-help that applies to everyone and helps no one.
How can I identify AI slop in text?
Look for these signals: opens with broad generalizations ('In today's fast-paced world...'), uses uncited statistics ('Studies show...'), sounds profound but says nothing specific, presents Wikipedia-level definitions as original insight, or offers universalized advice with no personal experience. The SlopDetector tool can automatically classify text into these 5 categories.
Can content be multiple types of slop at once?
Yes. Many pieces of slop combine multiple types. A LinkedIn post might open with Generic Slop ('In today's fast-paced world...'), transition to Pseudo-Insight ('The key is to find balance...'), and finish with Fake Authority ('Studies have shown that leaders who...'). Our detector identifies all patterns present in a single text.
Is all AI-generated content slop?
No. AI can produce substantive, well-cited, specific content when used as a tool rather than a replacement for thinking. The taxonomy classifies quality patterns, not origin. Humans write slop too — AI just makes it possible to produce it at unprecedented scale.
Test Your Content Against the Taxonomy
Our detector automatically classifies content into these categories and explains why.