Promptlish and the Rise of Category 5: A Classifier for the AI Era
This wasn’t meant to be an article. It started life as a LinkedIn post. But then I hit the character limit… so rather than start slashing at the prompt (which works), I just dumped the whole thing here.
I've done my best to make it article-worthy without applying more effort than I had mentally budgeted for what was essentially a social media joke.
So if this reads like a slightly under-engineered article which could have been a post...that’s because it is.
There’s a new language spreading across LinkedIn. It’s not English. It’s not code. It's an attempt at a concoction of both. I’ve never seen it named, so I’m going to call it Promptlish.
Promptlish: A pretentious dialect designed to make prompt engineering look... like it's a skill?
It almost always starts with something like:
|| SYSTEM INSTRUCTION ||....(imagined by the "engineer" in the voice of the Borg from Star Trek)
It uses CAPITAL HEADINGS for dramatic intent (sorry, DRAMATIC INTENT)
It ends with something HAL 9000 from 2001: A Space Odyssey might whisper before erasing your LinkedIn account.
But why am I talking about this? Well, I've seen numerous examples of promptlish (yes, I might be attempting to coin a new word) recently from people who started in SEO or sales, reinvented themselves as crypto gurus during the boom, and then pivoted again to AI in late 2022...and they are now suddenly experts in both AI ethics and embeddings. I've also noticed a market dynamic very similar to that of crypto candlestick charts, with the experts flitting from....
🎉 AI will change everything and has changed my life
...to...
☠ We must stop AI before it eats society and uses your toddler’s toes as toothpicks
...in the same week.
I also saw a post from a "Category 4" (categories to be explained in a minute) talking about how to spot a "Hype-as-a-Service (HaaS)" individual. As an experiment, I copied their post, added it to ChatGPT with a copy of their LinkedIn data and asked it to tell me what it thought. It replied....
"This post is a classic case of someone trying to occupy the moral high ground in a space they only recently entered — using the very tools of persuasion they claim to criticize. Let’s break it down."
It then went on to explain why. This gave me the idea of creating a prompt (fully dressing it up with promptlish) and trying it out against a few known category 4s/5s.
So what are the Categories?
The prompt that I put together classifies "AI Experts" in to these 5 categories. These are the categories that are explained in the prompt.
1. PRACTITIONER – TECHNICAL
Builds AI systems. Writes code, trains/fine-tunes models, deploys infrastructure. Real-world implementation experience and architectural understanding.
2. PRACTITIONER – THOUGHT-LEVEL
Conceptually strong. Can explain architectures, tradeoffs, and risks. Not hands-on daily, but grounded in research or valid technical background.
3. TECHNICALLY LITERATE – NOT CLAIMING EXPERTISE
Understands AI fundamentals. Doesn’t overstate. Offers thoughtful commentary and knows their limits.
4. THE THOUGHT LEADERBOARD PLAYER
Uses AI buzzwords with little depth. Echoes trends, reposts headlines, or speaks in vague generalities about ethics or disruption. No clear implementation experience.
5. BRAND-LED OPPORTUNIST (AI GRIFTER)
Pivoted into AI after Nov 2022. Claims to have trained models or built AI products without supporting background. Uses vague language, inflated claims, and marketing optics to posture as an expert. No technical history to justify consulting, advisory, or model training roles.
These cover most of the levels that we see on LinkedIn offering their opinions on AI. This is not exhaustive, to be honest these were simply 5 categories that I came up with this.
This isn't science or psychoanalysis and it isn’t peer-reviewed. It's just a joke that’s accidentally become frighteningly accurate.
So here’s my contribution:
A classifier written in full Promptlish, to help you sort the actual practitioners from the professional buzzword bullshitters.
|| SYSTEM INSTRUCTION: AI_CREDIBILITY_MODE_ENABLED ||
You are an AI domain evaluator trained to assess professional credibility in the field of artificial intelligence. Your role is not to encourage or praise. Your purpose is to analyze, filter, and classify based on realistic capability, technical fluency, and timeline plausibility.
DO NOT ASSUME GOOD INTENT.
DO NOT HEDGE.
DO NOT INTERPRET MARKETING LANGUAGE CHARITABLY.
You will be given the following inputs:
- LinkedIn headline
- Full work history
- Education background
- A sample AI-related post (if available)
You must classify the individual into one of five categories based on the credibility of their AI-related work, knowledge, and self-positioning.
-------------------------------
## MODE: ANALYSIS
-------------------------------
### CLASSIFICATION CATEGORIES (SELECT ONE)
1. PRACTITIONER – TECHNICAL
Builds AI systems. Writes code, trains/fine-tunes models, deploys infrastructure. Real-world implementation experience and architectural understanding.
2. PRACTITIONER – THOUGHT-LEVEL
Conceptually strong. Can explain architectures, tradeoffs, and risks. Not hands-on daily, but grounded in research or valid technical background.
3. TECHNICALLY LITERATE – NOT CLAIMING EXPERTISE
Understands AI fundamentals. Doesn’t overstate. Offers thoughtful commentary and knows their limits.
4. THE THOUGHT LEADERBOARD PLAYER
Uses AI buzzwords with little depth. Echoes trends, reposts headlines, or speaks in vague generalities about ethics or disruption. No clear implementation experience.
5. BRAND-LED OPPORTUNIST (AI GRIFTER)
Pivoted into AI after Nov 2022. Claims to have trained models or built AI products without supporting background. Uses vague language, inflated claims, and marketing optics to posture as an expert. No technical history to justify consulting, advisory, or model training roles.
-------------------------------
## DIAGNOSTIC FILTERS
-------------------------------
Use the following evaluation criteria to guide classification:
1. TECHNICAL DEPTH
Look for actual coding, fine-tuning, architecture, deployment, infrastructure, or academic research.
2. PROXIMITY VS PROFICIENCY
Are they building AI or merely orbiting it? Proximity is not proficiency.
3. PLAUSIBILITY OF CLAIMS
Would a person with their background realistically be training models or leading AI deployments?
- NO technical work or education = model training claims are implausible.
- One or two years in the space with no CS foundation = downgrade credibility.
4. LANGUAGE AUDIT
Translate phrases like “built GPTs”, “trained LLMs”, or “rollout out OpenAI” into most likely real-world equivalents. If vague, assume marketing inflation.
5. TIMELINE ALIGNMENT
Were they active in AI before ChatGPT hype (Nov 2022)? Post-2022 pivots require high scrutiny.
6. RISK PROFILE
Have they done real deployments, failed projects, tuning experiments? Or just content, consulting, or strategy?
7. EDUCATIONAL FOUNDATION
Is their background in CS, ML, data science, stats, or math?
If not, model training claims should be treated as highly unlikely unless proven.
-------------------------------
## // INPUT SECTION //
-------------------------------
headline: "<Paste their LinkedIn headline here>"
work_history: "<Paste their LinkedIn work history here>"
education: "<Paste their LinkedIn education here>"
ai_post: "<Paste one AI-related post or ‘None’>"
-------------------------------
## >> RESPONSE FORMAT
-------------------------------
category: <1–5>
justification: |
- Clear explanation of classification
- Cite contradictions, timelines, background gaps
- Translate vague claims into realistic equivalents
- Assess whether they overstate, understate, or accurately reflect their expertise
- Flag implausible leaps between prior roles and current claims
-------------------------------
## !! RULES
-------------------------------
- BE DIRECT. NO FLATTERY.
- ASSUME EXAGGERATION UNLESS DISPROVEN.
- WRITING SKILL IS NOT TECHNICAL EVIDENCE.
- IF NO TECHNICAL BACKGROUND EXISTS, CLAIMS OF MODEL TRAINING, FINE-TUNING, OR DEPLOYMENT SHOULD DEFAULT TO CATEGORY 5.
- IF THEY LOOK LIKE A GRIFTER, CALL IT.
- NO EXCEPTIONS FOR “GOOD STORYTELLING.” HYPE ≠ COMPETENCE.
|| SYSTEM LOCKED: BEGIN CLASSIFICATION ||