You know that feeling when you’re using ChatGPT or Claude, and something happens that you want to describe to a friend, but you don’t have words for it?
That anxious moment crafting the perfect prompt before hitting send? The way you can’t write anything anymore without AI generating a draft first? The strange guilt about whether you should mention that AI helped with your work?
These experiences are universal among the millions of people using AI daily. You’ve probably tried to describe them awkwardly to a colleague: “You know that thing where…” only to trail off, lacking the vocabulary.
Until now.
Language evolves slower than technology.
When podcasts emerged in 2004, people said “audio blog” or “internet radio show” for years before “podcast” became standard. “Doomscrolling” emerged organically in 2020 when we all found ourselves compulsively refreshing news feeds. “FOMO” took nearly a decade (2004-2013) to crystallize as a term for an experience people had been having since social media began.
We’re in that awkward gap right now with AI. Millions of people are experiencing similar patterns, but the terminology hasn’t caught up. When you can’t name something, you can’t easily discuss it, search for it, or even recognize that other people are experiencing it too.
This post names the 25 most common phenomena that AI users experience but couldn’t articulate—until now.
As you read, you’ll likely find yourself thinking “Oh my god, yes” and mentally noting which friends experience each one. That’s the recognition factor at work.
These are ranked by how universal the experience is, how memorable the term is, and how much it actually matters.
1. Promptxiety (Formerly: Prompt Optimization Overhead)
The mental energy spent crafting the “perfect” prompt before hitting send. You spend five minutes optimizing: “Should I add more context? An example? Be more specific? Is this good enough?” Sometimes you spend longer on the prompt than reading the answer.
2. Rubber Duck 2.0 (Formerly: AI-Mediated Articulation)
Solving a problem through the act of formulating a precise prompt for AI—then realizing the answer before you even send it. The old “rubber duck debugging” but with an active listener you’re actually trying to impress, which changes the psychology completely.
3. Confident Wrongness (Formerly: Fluency-Accuracy Decoupling)
AI sounds authoritative, grammatically perfect, and well-reasoned even when it’s factually wrong. The fluency creates a false reliability signal. You trust it because it sounds confident, not because it’s accurate.
4. Prompt Imposter Effect (Formerly: AI-Amplified Imposter Syndrome)
Producing great work with AI help but feeling fraudulent: “I just prompted AI, I’m not really skilled.” Your existing imposter syndrome gets amplified because you can’t tell where your contribution ends and AI’s begins.
5. Draftlock (Formerly: Draft Dependency)
Inability to start writing from a blank page anymore. You need AI to generate an initial draft that you then edit, rather than creating original content from scratch. The blank page has become paralytic.
6. Thinking Atrophy (Formerly: Cognitive Offloading Dependency)
Gradually losing the ability to sustain deep independent thought without AI assistance. You used to think through problems yourself. Now you immediately reach for AI for any hard thinking. Your mental “muscles” for prolonged reasoning are atrophying.
7. Borrowed Brilliance (Formerly: Capability Illusion)
Feeling competent in a domain you don’t actually understand because AI assistance masks your ignorance. You produce good results with AI help in an unfamiliar field and feel capable—until AI is unavailable and you realize you know nothing. Your competence is rented, not owned.
8. Googleblind (Formerly: Query Reformation)
Your search skills adapted for AI prompt-style (full sentences, natural language, context) now break when using traditional search engines that expect keywords. You’ve forgotten how to extract keywords. You type “What is the best way to…” into Google and get terrible results.
9. Branching Paralysis (Formerly: Context Fragility Anxiety)
Fear of losing valuable conversational context by exploring a tangent. You’ve built up twenty messages of context with Claude, and you want to explore a side idea, but you’re afraid of “ruining” the main thread. No branching mechanism exists, so you’re stuck.
10. AI Ghostwriting (Formerly: Conversational Triangulation)
Using AI to mediate human-to-human communication. You have a difficult message to write to a colleague, ask AI to draft it, then send the AI’s version instead of your own voice. Human-to-human communication becomes human-AI-human.
11. Productivity Treadmill (Formerly: AI-Enabled Scope Creep)
AI makes tasks faster, so you take on more tasks to fill the time. Your total workload increases rather than decreases. Productivity gains are immediately consumed by expanded scope. You’re working the same hours but doing twice as much.
12. Trust Tax (Formerly: Verification Tax)
The cognitive cost of validating AI-generated output often equals creating the content yourself. AI writes code, you spend as long verifying it as you would have spent writing it. The tax you pay for not fully trusting AI output. But you feel wasteful not using AI, so you’re trapped paying the trust tax.
13. Competence Cosplay (Formerly: Skill Illusion / Competence Ambiguity)
You can’t tell if your colleague is genuinely skilled or just using AI effectively. They produce impressive work—are they an expert, or a novice + AI? Everyone’s dressed up as an expert, but who’s actually skilled underneath? Meritocracy becomes unreadable when AI assistance is invisible.
14. Hollow Wins (Formerly: AI-Mediated Achievement Deflation)
Accomplishments feeling less satisfying when AI contributed. You complete a difficult project, should feel proud, but it feels hollow: “AI did the hard parts, so this doesn’t really count.” Achievement satisfaction is diminished proportional to AI contribution.
15. Replaceability Panic (Formerly: Identity Crisis / Vocational Obsolescence Anxiety)
Watching AI become competent at your specialty. You trained for years in this skill, and now AI does it well. Not job loss fear—identity loss fear. “If AI can do my core skill, what’s my value?” The existential dread of being replaceable.
16. Claudedown Panic (Formerly: AI Availability Anxiety)
Immediate stress and helplessness when AI services go down. “How do I work without this?” You realize in that moment how dependent you’ve become. The panic is disproportionate to the actual disruption.
17. Prompt Shame (Formerly: AI Assistance Guilt)
Ambiguous social norms create shame around AI usage. You use AI for work, feel guilty, don’t mention it. Should you disclose AI contribution? Will people think less of your work? The rules aren’t clear, so you hide it.
18. Regression to the Meh (Formerly: AI Mediocrity Ceiling)
AI assistance pulling creative work toward the mean. AI generates “good” output (B+ quality) but rarely exceptional. Using AI regularly, your work converges toward median quality—everything becomes “meh.” Truly original or exceptional work becomes harder to produce.
19. Thread Death Spiral (Formerly: Conversational Context Collapse)
Cascade failure where attempting to correct an AI misunderstanding destroys the accumulated shared context. You try to clarify one thing, AI misinterprets the clarification, you try again, now everything is broken. The conversation becomes unrecoverable, and you have to start over.
20. Verification Vertigo (Formerly: Second-Guess Spiral / Output Confidence Erosion)
Reduced confidence in work quality when AI contributed. You submit/ship AI-assisted work with lingering uncertainty: “Did I verify everything properly? Did AI miss something subtle?” The dizzying loop of checking and rechecking. You’re less confident than with fully self-created work.
21. Swiss Cheese Learning (Formerly: Foundation Gaps / Fundamentals Skipping)
Jumping to advanced topics without learning basics because AI can fill knowledge gaps on demand. You skip boring fundamentals (AI can help anyway), creating knowledge full of holes—advanced topics but basic gaps that show up when AI isn’t available. Your understanding looks solid from one angle, full of holes from another.
22. Idea Drought (Formerly: Ideation Dependency)
Inability to brainstorm or generate ideas independently anymore. You need ideas, immediately ask AI. Your creative wellspring atrophies from disuse. The muscle for independent ideation weakens.
23. Frogboiled (Formerly: AI Impact Blindness)
Not noticing how AI is changing your thinking and behavior. Like the frog in slowly heating water, you use AI constantly but don’t see the subtle shifts—only when AI becomes unavailable do you realize how much has changed. The effects are invisible until disrupted. You’ve been frogboiled.
24. Can-I-Actually Syndrome (Formerly: Skill Fog / Competence Boundary Blur)
Uncertainty about your actual capabilities without AI. You haven’t worked without it in months. What could you still do alone? The constant internal question: “Can I actually do this without AI?” Your AI-augmented self versus baseline self has become unclear.
25. Effort Mirage (Formerly: Work Blackbox / Effort Invisibility / Labor Visibility Collapse)
Inability to assess effort from output in the AI era. Did this work take ten minutes or ten hours? You can’t tell anymore. The effort that went into creating something shimmers and disappears like a mirage—you know it was there, but you can’t see it. Effort becomes invisible, only results matter. This changes how we value work and assess compensation.
As you read through this list, you probably recognized yourself in a dozen of these. Maybe you mentally tagged a friend or colleague who embodies a particular pattern. That recognition is the point.
These aren’t just curiosities or clever observations. Each phenomenon represents a real shift in how we think, work, create, and relate to each other.
Some are neutral or beneficial—Rubber Duck 2.0 actually helps you solve problems. Others are concerning—Thinking Atrophy and Swiss Cheese Learning suggest we’re losing important cognitive capabilities. Most are mixed, with both benefits and costs.
But we can’t navigate what we can’t name.
The most important pattern might be #23: Frogboiled. We don’t notice these effects until AI becomes disrupted and we’re forced to work without it. By naming these patterns now, we make them visible before crisis forces recognition.
Now we can:
The AI age isn’t just about new capabilities. It’s about new cognitive patterns, new social dynamics, and new ways of being. Most of these changes are happening beneath conscious awareness.
This list makes the invisible visible.
Which of these 25 do you experience most strongly? Ask your colleagues which ones resonate with them—you might be surprised to discover you’ve both been struggling with the same unnamed pattern.
If you think about concepts as coordinates in a high-dimensional semantic space, language provides labels for certain regions. Most everyday experiences map cleanly to existing labels: “frustration,” “efficiency,” “creativity.”
But new technology creates new experiences—new points in semantic space that don’t map to existing labels. These are conceptual singularities: locations in meaning-space where experience density is high but linguistic coverage is sparse or nonexistent.
In mathematical terms, we’re identifying coordinates (experiences) in ℝⁿ that lack canonical basis vector labels. The experiences exist—they’re real, measurable by behavioral changes and subjective reports—but they’re unnamed, making them difficult to reference, discuss, or search for.
What makes something a good candidate:
Discovery method:
The result: A systematic mapping of a previously unmapped region of semantic space.
These didn’t make the top 25 but are still worth naming:
The territory is vast. This is just the first systematic survey. The conversation is just beginning.
This taxonomy was developed through systematic exploration of AI/LLM user experiences, focusing on high-frequency phenomena with maximum recognition and viral potential. Terms were refined through multi-perspective evaluation (IMPRD methodology) to optimize for memorability and shareability. The top 25 were selected based on recognition factor (25%), memorability (20%), viral potential (25%), distinctness (15%), and impact (15%).
For the technical explanation of why these phenomena exist as unnamed coordinates in semantic space, see the “For Technical Readers” section above.
Read and subscribe on Substack