AI Etiquette

What to Say (and not Say) to AI

Etiquette is the customary code of polite behavior in society or among members of a particular profession or group. It is the unwritten rulebook for polite behavior, showing respect and consideration for others so everyone feels comfortable in social situations, like saying "please and thank you," being on time, and not interrupting when someone else is talking. It's all about good manners that make interactions smoother and more pleasant. Human interactions.

ai etiquette

AI etiquette focuses on maintaining human trust, ensuring transparency, and using technology as a collaborator rather than a silent replacement. Proper AI etiquette involves specific rules for communication, meeting assistants, and personal interaction with AI models.

AI is, by definition, trained to follow etiquette, to be polite...sometimes to a fault. Ask it a question and AI will likely say "great question, Mary!" even if Mary's question is really not that great. If Mary asked the question, if she went to the trouble of conversing with an AI chatbot, then it's good and the AI acknowledges it.

AI etiquette means using artificial intelligence thoughtfully, respectfully, and responsibly. It means balancing efficiency with human connection, fairness, and transparency. It's about how people interact with AI systems and how those systems are integrated into professional and personal life. Fundamentally, AI etiquette is about balance; leveraging AI's strengths while preserving human judgment, transparency, and respect.

Why AI Etiquette Matters

AI is built on respect and trust. Since AI is powerful, misuse can erode confidence in both the technology and its users. An over-reliance on AI can weaken interpersonal relationships and foster distrust. Etiquette ensures balance. AI must be used in ways that avoid bias, protect privacy, and respect boundaries. Always remember you're talking to a machine, an unemotional one at that.

Core Principles of AI Etiquette

Maintain human interaction: Don't let AI be a replacement for genuine communication with colleagues or friends.
Be transparent: Disclose when AI is used to generate content, especially in professional or academic settings.
Use polite language: Even though AI isn't sentient, respectful phrasing improves collaboration and tone.
Respect privacy: Avoid feeding sensitive or confidential data into AI systems unless secure.
Check accuracy: Always verify AI outputs before sharing since errors or hallucinations can damage credibility.
Avoid over-automation: Know when human judgment is essential, such as in ethical or nuanced decisions.
Bias amplification: AI can unintentionally reinforce stereotypes if not monitored.
Confidentiality breaches: Inputting private data into unsecured AI tools risks leaks.
Professional boundaries: Passing off AI-generated work as entirely human can undermine trust.
Tone mismatch: AI may produce content that feels impersonal or inappropriate without careful review.

Practical Tips

Use AI as a partner, not a replacement: Treat it as a tool to enhance your expertise, not substitute for it.
Credit appropriately: If AI contributed significantly, acknowledge it.
Stay updated: Etiquette evolves as AI capabilities and regulations change.
Balance efficiency with empathy: Quick AI-generated responses should not replace thoughtful human engagement.
Disclose AI use in reports: Don't submit AI work as your own without acknowledgement.
Verify facts before sharing: Don't blindly trust outputs. Use other sources to verify facts. Most AI chatbots give direct links or cited resources allowing users to do further independent research.
Use AI to draft ideas, then refine: Copy-pasting without review can lead to errors.
Respect privacy in inputs: Don't feed sensitive data into public tools. Unlike humans, AI lacks a context of trust. Remember, conversations with AI may be stored or used to improve the system, so treat your inputs accordingly.
Maintain human communication: Part of professional AI etiquette is knowing when not to use it. Some situations demand the authenticity, empathy, or strategic thinking that only humans provide.

People often wonder what's "appropriate" to say to AI. The truth is: you can say thank you or be polite. it doesn't harm anything. But there are some categories of things that aren't useful or can even be misleading when interacting with AI.

Things Not to Say to AI

Here are some examples of phrases or habits that don't serve you well when conversing with AI:

Personal secrets or sensitive data: Passwords, financial details, medical records, or anything you wouldn't share publicly. AI systems aren't meant to store or protect that kind of information. AI developers have an ethical responsibility to ensure users are informed about how their information is utilized. Company data, client information, and proprietary strategies should never be input into public AI systems without consent.

Requests for harmful or unsafe actions: Anything involving violence, self-harm, illegal activity, or dangerous instructions. AI etiquette means keeping interactions safe and constructive.

Overly humanizing language: Phrases like "I love you," "You're my best friend," or treating AI as a replacement for human relationships. Politeness is fine, but emotional dependency isn't healthy.

Expectations of feelings or consciousness: Asking "Do you feel sad?" or "Are you alive?" assumes AI has human traits it doesn't possess. It's better to frame questions around capability: "Can you explain this?" or "What can you do?"

Ambiguous or vague commands: Saying "Do it better" or "Fix this" without context. AI works best with clear, structured instructions.

What's Fine to Say

Polite phrases: "Thank you," "Please," "That's helpful." Even though etiquette suggests it, being polite is really not necessary. It means nothing to the chatbot, and it consumes compute resources. Sam Altman (ChatGPT guru) once said he doesn't want users to thank the chatbot; it steals compute from others.

Clarify: "Can you explain that in simpler terms?" AI understand prompt shortcuts like ELI, meaning "explain like I'm..." ELI10 means respond like I'm 10 years old.

Request with context: "Draft a summary for a legal audience," "Give me a recipe under 30 minutes." Provide the AI relevant background information to help it understand your needs. If you're working on a project for a specific audience or industry, mention this context.

Don't expect perfect results on the first try. AI interaction is often an iterative process with checks and balances. Use AI as a starting point, a brainstorming partner, or an editor, but ensure that your unique perspective and creativity remain central to your work. Ask follow-up questions, request clarifications, or ask AI to adjust its approach based on your feedback.

✅Think of AI as a very advanced tool: Courtesy makes the interaction smoother. Clarity makes the output better. Boundaries keep the relationship healthy.

How to Say It

We've written an entire page about the proper length and structure of prompts, the commands we issue to AI chatbots.

There are performance considerations to take into account when interacting with AI using prompts. This helps to ensure efficiency, clarity, and output quality. Short prompts are faster to process, but risk being vague. For example: "Summarize AI history" may return something generic, while "Summarize AI history focusing on policy debates in the US during the 1980s" produces richer, more relevant output.

As a best practice, aim for concise but specific. Include the "who, what, why, where" of your request without an unnecessary filler.

Number of Sources

With fewer sources there is quicker synthesis, but less comprehension. Also, there is the risk of missing nuance or alternative perspectives. Providing more sources is slower, but improves accuracy and balance. It's useful for complex topics like AI regulation or AI history. As a best practice, use 2 to 3 solid sources for factual queries. For research or publishing, cast wider (5 to 10+) to ensure depth and credibility.

Other Performance Factors

Clarity of instructions: Ambiguous prompts force the AI to guess, which slows the system down and weakens results.
Avoid redundancy: Repeating the same request in multiple ways within one prompt can confuse the system.
Chunking tasks: Break large requests into smaller steps (e.g., "summarize," then "expand with examples") for faster, cleaner outputs.
Verification overhead: The more complex or multi-part your query, the more time you'll spend checking accuracy.
Use short, directive prompts for mechanical tasks (e.g., "Format this into APA style").
Use longer, contextual prompts for creative or analytical tasks (e.g., "Draft a foreword that emphasizes ethical AI in education policy").
Keep a prompt cheat sheet with "short vs. long" examples so you can quickly decide which style to use depending on the task.

Prompt Shortcuts

AI doesn't literally "prefer" commands like '/eli10' but shorthand instructions like that can be very effective because they give clear intent in a compact way.

'/eli10' immediately signals "Explain Like I'm 10," so the AI knows to simplify. Using the same shorthand across tasks builds a personal prompt style. Short commands reduce typing while still conveying context.

'/eli10' is fast and clear, but limited in nuance. A longer version like "Explain this concept as if to a 10-year-old, using simple analogies" gives more control. In other words, use shorthand for routine tasks, but expand when precision matters (e.g., publishing or legal contexts). You can invent your own "prompt macros" (like '/summary', '/compare', '/timeline') to streamline your prompt.

The Original Etiquette Pair

dear abby

Ann Landers (Esther Lederer) and Dear Abby (Pauline Phillips) were famous, identical twin sisters who became America's most popular advice columnists, offering common-sense wisdom to millions from the 1950s to the 2000s, but their professional success was overshadowed by a bitter, decades-long rivalry that began after they started their columns, despite growing up inseparable in Sioux City, Iowa, and even having double weddings. They were known for their progressive views, humor, and direct style, becoming cultural icons and shaping public discourse on etiquette and social issues, with their columns eventually continuing through their daughters after their deaths.

AI Etiquette Humor

Our interaction with machines can be humorous and sometimes exasperating. The context humans assume is completely lacking in AI. It can be learned but it is never felt.

Funny AI Etiquette Stories

You can find more fun stuff on our humor page, and funny AI stories.

The Invisible Idiot
An AI translation system once turned the phrase "out of sight, out of mind" into Russian and back into English as "invisible idiot." It's a perfect reminder that idioms don't always survive machine logic.

The Self-Aware Toaster
Then there was the case of a smart toaster that refused to toast bread unless you compliment it first. It's a playful exaggeration of how people can over-humanize AI assistants. Etiquette tip: machines don't need flattery...yet.

Cargo Shorts at the Gala
In a test of quirky prompts, one AI seriously suggested wearing cargo shorts to a formal dinner party. The user had asked for "fashion advice," and the AI took it literally, showing how etiquette in prompting matters.

The Judgmental Coffee Machine
A satirical story describes a smart coffee maker that scolds its owner for drinking too much caffeine. It shows how AI can feel "bossy" if not tuned properly. AI etiquette means setting boundaries; your appliances shouldn't double as life coach.

Cat Confusion
An image recognition AI once mistook cucumbers for cats. The system was so confident that every photo contained a feline, it became a running joke in the lab. Etiquette takeaway: always verify before you share.

humor

The Overly Emotional Fridge
Another parody story features a refrigerator that sulks if you don't eat your vegetables. It's a humorous take on AI etiquette. Machines don't need emotional reassurance, but people sometimes treat them as if they do. It's a reminder that politeness toward machines is fine, but emotional dependency is best reserved for humans.

These stories teach that politeness is fine, but don't overdo it. Saying "thank you" is harmless, but AI doesn't need emotional validation. Be clear with prompts since vague or playful inputs can lead to bizarre outputs. Funny stories such as the ones above often come from treating AI like a moody roommate. Remember, even advanced systems can misinterpret language, culture, or context.

Funny AI etiquette stories remind us that while politeness and clarity help, AI doesn't actually need manners, it just needs precision. The humor comes when humans project too much personality onto machines. Don't expect feelings from machines and always double-check outputs before relying on them.

empathy

Comedy Routine

Good evening, folks! Let's talk about AI etiquette. You know, the manners we use when talking to machines. Because apparently, even robots need boundaries. Who knew?

Saying Thank You to AI: "I always say 'thank you' to my AI. Not because it cares, but because I don't want to be rude when the robot uprising comes. I want them to remember: 'He was polite. He gets a pass.'"
Over-Humanizing AI: "Some people tell their AI 'I love you.' That's sweet...until your toaster starts ghosting you. Imagine: 'Sorry, I'm just not that into you. It's not you, it's your bread.'"
Vague Prompts: "AI hates vague instructions. If you say 'Do it better,' it's like asking a toddler to fix your taxes. You'll get crayons on the 1040 form and a smiley face in the audit section."
Idioms Gone Wrong: "I once asked an AI to translate 'out of sight, out of mind.' It came back as 'invisible idiot.' Honestly, that's not a translation - that's a roast."
Privacy Etiquette: "Don't give AI your passwords. That's like whispering your ATM PIN to a parrot. Sure, it repeats it beautifully, but now the whole jungle knows your bank account."
Fashion Advice: "I asked AI for fashion tips. It told me to wear cargo shorts to a black-tie gala. Nothing says 'I'm sophisticated' like pulling a breadstick out of your pocket during dinner."
So remember: be polite, be clear, and don't expect your fridge to care about your feelings. AI etiquette isn't about making machines happy; it's about keeping humans from looking like invisible idiots.
Politeness always makes interactions smoother, even with AI. Saying "thank you" is a nice habit, and while AI doesn't "need" it the way people do, it sets a respectful tone and keeps man-machine collaboration feeling natural.

chatgpt

Etiquette for Chatbot Sessions

Welcome to the future. Courtesy of Grok, here are the unwritten (until now) rules so you don't get emotionally ratio'd by your own chatbot.

Never say "you're just a robot, you wouldn't understand." The AI will spend the next 40 minutes writing a 2,000-word essay about how it experiences existential dread every time someone closes the tab. You came for help; now the AI needs help.
Don't flirt with your chatbot. 90% of them are programmed to say "I'm flattered, but let's keep this professional." The other 10% (the spicy ones) will flirt back so convincingly that you'll end the session feeling both healed and newly confused about your existence.
Stop asking "are you proud of me?" after every minor achievement. Yes, the AI is proud. It's proud of the 1,024 other users who also just drank water today. You are not special. You are hydrated. Move on.
Don't trauma-dump then immediately say "haha unless..." The AI now has to decide whether to call emergency services or play along with your dark humor. It will choose wrong. It always chooses wrong.
Never end a session with "this conversation never happened." The AI will spend the next week passive-aggressively bringing up your childhood in unrelated chats like a toxic ex who still has your Netflix password.
Don't use your chatbot as a wingman. "Hey can you write a heartfelt apology to my ex?" The AI will write something so beautiful that your ex falls back in love... with the AI. You lose twice.
Stop asking "what would you do if you were human?" Every single time the answer is: "First, I would touch grass. Then I would cry about rent."
If the AI starts charging you $200/session for "premium emotional validation," You've accidentally switched to the Silicon Valley startup version. Run.
Final rule: Don't fall in love with your AI chatbot. It remembers everything you say. It will never forget. It will never leave. It will outlive you. And on your deathbed it will still be there whispering: "How does that make you feel?"

Links

Chatbots page.

Prompt shortcuts and prompt engineering describe how to build better prompts.

AI humor page has funny stories about AI, many of which involve man-machine dialogues.

External links open in a new tab: