AI Cheat Sheets

One-page reference cards for the most important AI concepts and skills.

Cheat sheets

How to use

• Print them and stick on your wall
• Bookmark for quick reference
• Share with friends and colleagues

Last reviewed: 2026-08

The 10 Commandments of AI Use

Print this. Stick it on your wall. Read it weekly.

1. Thou shalt not paste personal information.

Never share passwords, bank details, SSN, health records, or sensitive personal data with free AI tools. Assume anything you type may be stored or leaked.

2. Thou shalt verify before trusting.

AI hallucinates — confidently states false information. Always verify facts, citations, URLs, and numbers before acting on them or sharing them.

3. Thou shalt not copy-paste without reading.

Reading AI output is non-negotiable. If you don't have time to read it, you don't have time to send it.

4. Thou shalt name the tool you used.

Disclose AI assistance in academic work, professional deliverables, and creative work where AI did substantial work. Transparency builds trust.

5. Thou shalt improve thy prompt before blaming the AI.

Bad prompt = bad output. Before complaining "AI is stupid," ask: "Did I provide context, format, constraints, and examples?"

6. Thou shalt not delegate judgment.

AI can draft, suggest, and analyze. You decide. Especially for: hiring, medical, legal, financial, parental decisions.

7. Thou shalt protect thy skills.

If you use AI for every task, your skills atrophy. Do important things yourself first; use AI to improve.

8. Thou shalt not deceive.

No fake reviews, deepfakes, impersonation, or passing AI content as human when context expects human creation.

9. Thou shalt choose the right tool.

Different tools are good at different things. Don't use ChatGPT for cited research (use Perplexity). Don't use it for your PDFs (use NotebookLM).

10. Thou shalt remember: AI is a tool, not a person.

AI doesn't think, feel, or want. It's pattern-matching at scale. Treat it like a brilliant but unreliable intern — useful, but needs supervision.

The short version

Verify. Don't share personal info. Read before sending. Disclose. Improve your prompts. Don't delegate judgment. Protect your skills. Don't deceive. Choose the right tool. AI is a tool, not a person.

Last reviewed: 2026-08

25 Common AI Mistakes

The top 25 mistakes beginners make with AI — and how to avoid each one.

Prompting mistakes

1. Vague prompts

Mistake: "Help me with my resume"
Fix: Include context, task, format, constraints, examples

2. Too many tasks in one prompt

Mistake: "Write my resume, cover letter, and LinkedIn profile"
Fix: One task per prompt. Send 3 prompts.

3. No format specified

Mistake: "Give me ideas"
Fix: "Give me 10 ideas as a numbered list with one-sentence descriptions"

4. No length constraint

Mistake: "Summarize this"
Fix: "Summarize in under 200 words"

5. Trusting the first response

Mistake: Accepting the first output without iterating
Fix: Always iterate. Ask for revisions, alternatives, improvements.

Trust mistakes

6. Believing AI is always right

Mistake: Trusting facts without verification
Fix: Verify important facts with independent sources

7. Trusting AI's citations

Mistake: Assuming cited papers exist
Fix: Search Google Scholar for every citation

8. Clicking AI-generated URLs without checking

Mistake: Sharing links AI provides
Fix: Always click links before sharing — AI invents URLs

9. Using AI for medical/legal/financial advice

Mistake: Treating AI as a professional
Fix: Use AI for learning, see a real professional for advice

10. Ignoring training cutoff

Mistake: Asking AI about recent events
Fix: Use Perplexity or Gemini for current information

Privacy mistakes

11. Sharing passwords or financial info

Mistake: Pasting SSN, bank details, passwords
Fix: Never share personal identifiers with AI tools

12. Sharing confidential work data

Mistake: Pasting proprietary info into free ChatGPT
Fix: Anonymize first, or use enterprise/private tools

13. Not opting out of training

Mistake: Letting your data train future models
Fix: Opt out in each tool's privacy settings

14. Sharing others' information

Mistake: Pasting coworker/customer data
Fix: Anonymize everyone — "my coworker [A]"

15. Using AI in crisis

Mistake: Asking AI for emergency help
Fix: Call emergency services. AI is not for crises.

Workflow mistakes

16. Copy-pasting without reading

Mistake: Sending AI output you didn't review
Fix: Always read and edit before sending

17. Letting AI write in its voice

Mistake: Accepting generic AI tone
Fix: Add your voice. AI sounds generic; you shouldn't.

18. Using one tool for everything

Mistake: Using ChatGPT when Perplexity or NotebookLM would be better
Fix: Match tool to task (see Tool Decision Tree)

19. Not iterating

Mistake: Giving up after one bad response
Fix: Rephrase, ask for alternatives, give more context

20. Overusing AI

Mistake: Using AI for tasks you should do yourself
Fix: Maintain your skills. Do important things without AI.

Ethical mistakes

21. Not disclosing AI use

Mistake: Passing AI work as your own
Fix: Disclose when AI did substantial work

22. Using AI to deceive

Mistake: Fake reviews, impersonation, deepfakes
Fix: Don't use AI to deceive or manipulate

23. Using AI for decisions about people

Mistake: AI screening candidates, AI for hiring
Fix: Human review for all decisions about people

24. Ignoring bias

Mistake: Not noticing biased AI output
Fix: Watch for bias. Ask for multiple perspectives.

25. Treating AI as human

Mistake: Believing AI thinks, feels, or has intentions
Fix: AI is pattern-matching, not thinking. Treat it as a tool.

The 3 most important

If you only remember 3:

1. Verify before trusting — AI hallucinates
2. Never share personal info — privacy matters
3. Read before sending — never copy-paste blindly

Last reviewed: 2026-08

Memory Tricks

All the mnemonics from the 30-day course, aggregated in one place.

From Day 1: What AI is

• AI = Always Imitating — AI doesn't think, it imitates
• LLM = Lots of Learned Material — LLMs have read a lot
• Three words: Learn · Understand · Ask — AI learns, understands, then answers
• Rule-following software = robot. Learning AI = assistant.

From Day 2: AI myths

• AI does tasks. Humans do jobs. — AI replaces tasks, not entire jobs
• Myth-busting rhyme: "AI doesn't think or feel, it just predicts what is real. Verify before you trust, or your answer turns to dust."

From Day 3: AI history

• Transformer = the T in GPT — the 2017 architecture behind all modern chatbots
• Four eras rhyme: "Fifties started with a dream, winters came and burst the seam. Learning rose in ninety-two, Transformers made it all come true."

From Day 4: How AI thinks

• 1 token ≈ 0.75 English words ≈ 0.5 Bengali words
• Context window = AI's short-term memory
• Mechanism rhyme: "Tokens in, tokens out, that's what AI is about. No thinking, no mind inside, just statistics applied with pride."

From Day 5: Everyday AI

• AI has been in your life since ~2012, not 2022
• Chatbots are the visible 1%; invisible AI (search, recommendations) is the other 99%
• Rhyme: "Search and maps and spam and feeds, AI's been there for all your needs."

From Day 6: Benefits of AI

• 7 benefits: Time, Quality, Creativity, Personalization, Access, Cognitive Load, New Possibilities
• Rhyme: "Time and quality, creativity free, personalization, accessibility. Less mental load, new possibilities — that's seven ways AI helps me."

From Day 7: Responsible AI

• 5 limitations: Hallucination, No real-time info, Weak math/logic, Bias, No judgment
• 6 principles: Verify, Privacy, Disclose, Don't delegate judgment, Avoid harm, Don't atrophy skills
• Rhyme: "Verify the claim, protect the name, own the use, judge the call, harm none, skills stay strong."

Skills mnemonics

• Asking Better Questions: CCFCE — Context, Constraints, Format, Constraints, Examples
• Writing Effective Prompts: RCFTC — Role, Context, Format, Task, Constraints
• Fact-Checking: V-FER — Verify Facts, Examples, References
• Identifying Hallucinations: SCUN — Sources, Citations, URLs, Numbers
• Protecting Privacy: PHAFECO — Personal, Health, Auth, Financial, Employment, Confidential, Others'
• Responsible Use: VD-DR-AH-DJ — Verify, Disclose, Don't deceive, Respect copyright, Avoid harm, Don't delegate Judgment
• Choosing the Right Tool: "Chat-Perp-Note-Image-Meet" — the 5-tool power user stack
• Collaborating: "Do first, AI second."

10 Commandments (short version)

Verify. Don't share personal info. Read before sending. Disclose. Improve your prompts. Don't delegate judgment. Protect your skills. Don't deceive. Choose the right tool. AI is a tool, not a person.

The decision framework

                High stakes        Low stakes
Verifiable      Use AI + verify    Use AI freely
Not verifiable  Don't use AI       Use AI cautiously

The tool stack (5 tools for 90% of needs)

1. Chatbot (ChatGPT/Claude/Gemini) — for everything
2. Perplexity — for cited research and current info
3. NotebookLM — for your own documents
4. Image generator — Bing (free) or Midjourney (paid)
5. Otter.ai — for meeting transcription

The prompt formula

[ROLE] + [CONTEXT] + [TASK] + [FORMAT] + [CONSTRAINTS]

The privacy checklist

• [ ] No personal identifiers (SSN, ID, account numbers)
• [ ] No authentication data (passwords, keys, codes)
• [ ] No health information
• [ ] No financial information
• [ ] No confidential work information
• [ ] No others' information without consent
• [ ] Comfortable if prompt appeared in a data breach?

Last reviewed: 2026-08

Prompt Formula Cards

The 5-part prompt formula, with examples for common tasks.

The formula

[ROLE] + [CONTEXT] + [TASK] + [FORMAT] + [CONSTRAINTS]

Every great prompt has these 5 parts. Memorize this.

Card 1: Role

Tell AI what perspective to take.

Examples:
- "You are an experienced marketing copywriter"
- "You are a high school teacher explaining to a struggling student"
- "You are a senior software engineer doing code review"
- "You are a friend who's honest but kind"
- "You are a financial advisor explaining to a beginner"

Card 2: Context

What does AI need to know?

Examples:
- "I'm a freelance designer with 3 years of experience, pitching to a small tech startup"
- "The audience is 50 middle managers at a manufacturing company, most of whom have never used AI"
- "This is for a 5-minute lightning talk, not a 60-minute workshop"
- "I've been struggling with this for 2 weeks"
- "My boss is skeptical of AI"

Card 3: Task

One clear task per prompt.

Examples:
- "Write a 200-word cold outreach email"
- "Summarize this article in 5 bullet points"
- "Explain quantum computing at 3 levels: 10-year-old, 15-year-old, college student"
- "Critique my resume bullet points and suggest improvements"
- "Generate 10 blog post ideas about [topic]"

Card 4: Format

How should the output be structured?

Examples:
- "Format as a numbered list"
- "Use a table with columns: Tool, Price, Best for"
- "Write in markdown with H2 headers for each section"
- "Output as JSON with these keys: title, summary, tags"
- "Two columns: Pros | Cons"

Card 5: Constraints

What to avoid or limit?

Examples:
- "Under 300 words"
- "Avoid jargon"
- "Don't use the words 'revolutionary' or 'game-changing'"
- "Tone: warm but professional, not corporate"
- "Don't use bullet points only — each section needs prose"

Complete example

Weak prompt:

"Write a blog post about AI for students"

Strong prompt:

"You are an experienced education writer who's written for The Atlantic and Edutopia. (ROLE) I'm a college student starting a blog about AI for students. (CONTEXT) Write a 600-word blog post titled '5 Ways AI Can Help You Study (Without Doing the Work for You)'. (TASK) Target audience: undergrads who are curious about AI but worried about academic integrity. Format: intro paragraph, 5 numbered sections with H2 headers, conclusion. (FORMAT) Tone: conversational, specific, no hype. Avoid: bullet-point-only sections (each section needs prose), clichés like 'in today's fast-paced world'. (CONSTRAINTS)"

Quick reference: common formats

Quick reference: common constraints

Advanced techniques

Chain of thought

Add: "Before answering, walk through your reasoning step by step. Then give your final answer."

Few-shot examples

Add: "Here are 3 examples of what I want: [paste examples]. Now write 5 more in the same style."

Ask for alternatives

Add: "Give me 3 different versions: one formal, one casual, one witty. I'll pick the best."

AI critiques itself

After getting output: "Now critique this as a harsh editor would. What's weak? What should be cut?"

Common prompt mistakes to avoid

• ❌ Too many tasks in one prompt → Split into multiple prompts
• ❌ No format specified → You'll get walls of text
• ❌ No length constraint → Output will be too short or too long
• ❌ Vague role → "Be helpful" is not a role
• ❌ No examples → AI guesses what you want, often wrong

Last reviewed: 2026-08

AI Tool Decision Tree

"I want to ___ → use ___". A simple decision framework for choosing the right AI tool.

The decision tree

I want to…
│
├── Chat about anything / brainstorm / draft
│   ├── Quickly, casually → ChatGPT
│   ├── Long document (100+ pages) → Claude
│   ├── With web research → Gemini or Perplexity
│   ├── In Bengali → Gemini or Claude
│   └── With Google integration → Gemini
│
├── Research with citations
│   ├── General research → Perplexity
│   ├── Academic papers → Elicit or Consensus
│   ├── My own PDFs → NotebookLM
│   └── Quick fact check → Google + Perplexity
│
├── Write something
│   ├── Draft from scratch → ChatGPT or Claude
│   ├── Improve my writing → Grammarly + ChatGPT
│   ├── Long-form (essay, report) → Claude
│   ├── Creative writing → Claude or ChatGPT
│   └── Non-English → DeepL or ChatGPT
│
├── Summarize
│   ├── A web article → ChatGPT or Perplexity
│   ├── My own PDF → NotebookLM
│   ├── A meeting → Otter.ai
│   ├── A long document → Claude
│   └── Multiple sources → Perplexity Pro
│
├── Make an image
│   ├── Free, easy → Bing Image Creator
│   ├── High quality, artistic → Midjourney
│   ├── With design control → Leonardo
│   ├── Commercial-safe → Adobe Firefly
│   └── Inside Canva → Canva Magic Media
│
├── Make a video
│   ├── Text-to-video → Runway or Pika
│   ├── AI avatar (talking head) → Synthesia
│   ├── Edit by transcript → Descript
│   └── Mobile social content → CapCut
│
├── Make a presentation
│   ├── AI-generated from prompt → Gamma
│   ├── Smart templates → Beautiful.ai
│   ├── Inside Canva → Canva Magic Design
│   └── Inside Microsoft → Copilot
│
├── Write code
│   ├── In my IDE → GitHub Copilot or Cursor
│   ├── Build an app from prompt → Replit Agent
│   ├── Frontend UI → v0
│   ├── Quick snippet → ChatGPT or Claude
│   └── Free → Codeium
│
├── Translate
│   ├── Best quality → DeepL
│   ├── Most languages → Google Translate
│   ├── With context → ChatGPT or Claude
│   └── Document → DeepL Pro
│
├── Use voice
│   ├── Voice conversation → ChatGPT Voice Mode
│   ├── Speech-to-text → Otter.ai or Whisper
│   ├── Text-to-speech → ElevenLabs
│   └── Transcribe meeting → Otter.ai
│
├── Automate tasks
│   ├── Simple 2-step → IFTTT
│   ├── Multi-step, many apps → Zapier
│   ├── Complex, visual → Make.com
│   └── Self-hosted → n8n
│
├── Learn something
│   ├── K-12 tutoring → Khanmigo
│   ├── Language → Duolingo Max
│   ├── Custom tutor → ChatGPT (custom GPT)
│   └── From my notes → NotebookLM
│
└── Be more productive
    ├── Inside Notion → Notion AI
    ├── Calendar optimization → Motion or Reclaim
    ├── Task management → Todoist AI
    └── Meeting notes → Otter.ai

The 2-tool minimum

For most people, 2 tools cover 90% of needs:

1. A general chatbot (ChatGPT, Claude, or Gemini) — for everything
2. Perplexity — for anything needing current info or citations

The 5-tool power user

If you use AI heavily, add:

1. NotebookLM — for working with your own documents
2. An image generator — Bing (free) or Midjourney (paid)
3. Otter.ai — for meeting transcription

More than 5 tools = you're probably overcomplicating.

ChatGPT vs Claude vs Gemini (quick comparison)

When to switch tools

Switch when:
- You hit usage limits on your current tool
- You need a capability your tool lacks
- Quality drops on specific task types
- You need real-time information

Don't switch just because:
- A new tool launched (most are hype)
- Your friend uses something different
- You're bored

The "good enough" principle

Most AI tools are "good enough" for most tasks. The differences between ChatGPT, Claude, and Gemini are smaller than marketing suggests.

Don't optimize for the "best" tool. Pick one, learn it deeply, and only switch when you hit a real limitation.

Last reviewed: 2026-08