Don't paste the AI, please
TL;DR Highlight
A satirical site says read the AI's answer and reply in your own words instead of pasting it.
Who Should Read
Developers who tend to paste ChatGPT answers straight into colleagues' Slack or code review questions
Core Mechanics
- The asker wants your answer
- The asker has the same AI tools you do.
- They asked for your context, taste, and judgment.
- Use AI as a drafting tool
- Read what it gave you, then write your own version or polish it.
- Don't be a middleman between the AI and the answer.
- Pick only the part that answers
- Ignore the rest; three sentences is enough.
- Even copied sentences are better if you actually read them.
- Quote useful bits and say why
- Example: "I asked Claude and this bit here makes sense:".
- If you have nothing to add, say "No strong opinion here" or stay silent.
- It's satire with variants
- The site calls itself satire and allows copying, translating, and changing.
Community Reactions
- Pasting shifts the burden of understanding to others, one commenter argued
- An AI enablement firm's principles doc starts with "Write as yourself".
- Slack and email should be self-written, while long deliverables may use agents.
- Long AI messages beat context-free ones, one commenter said
- Colleagues' messages like "x broken" lack context, while Claude-written ones give full context.
- They use it as a resume point, then continue as humans.
- Validation is the missing step, one commenter said
- Accuracy and relevance must be checked between reading and rewriting.
- Skipping it makes you a "human ai beautifier".
- Commenter's boss pastes AI code reviews
- They review every AI item themselves, which takes 3 hours.
- The boss still angrily asks why it isn't done.
How to Apply
- Apply a "three sentences I actually read" rule today
- Keep only the parts of the AI answer that answer the question.
- Verify the facts yourself before sending.
- Cite the source and reason when using AI
- Quote it like "I asked Claude and this bit makes sense:".
- Add one line on why you agree.
- Add AI usage principles to team guidelines
- Specify that Slack and email are self-written while long deliverables may use AI.
- State that the author owns the output.
Try It
# Paste into Claude or ChatGPT: a lite version of the article's method (read, then write yourself)
Below is a colleague's question and the AI answer I got.
Question: {{question}}
AI answer: {{AI answer}}
Pick only the parts that actually answer the question and cut to 3 sentences or fewer. List separately any claims I should verify.
Next: verify the list yourself, then rewrite the answer in your own voice and send it.Terminology
Related Papers
AI is a tool and using it well takes skill
A rebuttal to Hickey: LLMs are tools, and skill in directing and checking them decides the outcome
Reflective Prompt Optimization of a Deep-Research Agent: A Topic-Disjoint, Multi-Seed Protocol for Measuring Transferable Gains and Prompt-Level Overfitting
Reflective prompt optimization raised held-out citation reward from 0.702 to 0.781 on a deep-research agent.
GPT‑6 and Intelligent UI for everyone
GPT-6 reportedly adds generated UI to chat; with no article text, only comment reactions are available.
Base Models Can Reason By Taking a Cue From Training Data
Prefilling just two opening tokens lifted Olmo-3-7B's MATH-500 accuracy from 42% to 78%.
Labels Override Definitions in Jev-Style Typed Decision Models
Renaming options to A/B raised routing accuracy by +0.1511 over meaningful labels.
How Much Prompt Is Enough? A Blackbox Minimization of Few-Shots in LLMs
Delta debugging trimmed few-shot prompts by 65.3% on average with unchanged propositional output