Working with AI

Core Collaboration Skills

How to Write Effective Prompts

Good prompts are clear, specific, and give the AI enough context to help you effectively. Think of prompting like giving instructions to a skilled colleague who doesn't know your project history.

Key principles for effective prompts:

• Provide context: Share relevant background information
• Be specific about format: HTML? Markdown? Code? Tell the AI what you want
• State your constraints: Length limits, style requirements, audience level
• Show examples: "Like this, not that" is incredibly powerful

Iterative Refinement: Ask, Review, Adjust

The best results come from iteration, not perfection on the first try. Think of AI collaboration as a conversation, not a vending machine where you insert a prompt and get a final product.

The refinement cycle:

1. Ask: Make your initial request with context
2. Review: Examine what the AI produces critically
3. Adjust: Give specific feedback on what to change
4. Repeat: Keep refining until it meets your standards

Don't accept the first response if it's not quite right. The AI can't read your mind, but it can respond to clear feedback about what's working and what isn't.

When to Be Specific vs. When to Be Open-Ended

Knowing when to constrain the AI and when to let it explore is a critical skill.

Be specific when:

• You know exactly what format you need (HTML structure, specific file types)
• Technical accuracy is critical (code that must work a certain way)
• You're working within strict constraints (character limits, specific requirements)
• You want consistency with existing work (matching a style you've established)

Be open-ended when:

• You're brainstorming or exploring possibilities
• You want creative approaches to a problem
• You're not sure what the best solution looks like
• You want the AI to suggest alternatives you haven't considered

Building Context Across Conversations

AI doesn't automatically remember your previous conversations, but you can build continuity through deliberate context management.

Effective context-building techniques:

• Session summaries: Start new conversations with a summary of previous work
• Upload reference files: Share relevant documents, code, or examples
• Reference previous decisions: "As we discussed earlier..." helps maintain continuity
• Keep project documentation: Maintain notes about decisions, standards, and approaches

Practical Techniques

Breaking Big Tasks into Smaller Pieces

Large projects can overwhelm both humans and AI. Break them down into manageable chunks that can be completed, reviewed, and refined.

Effective task breakdown:

• One file at a time: Don't ask for an entire website; build it page by page
• One section at a time: Complex documents work better in sections
• Prototype first: Start with a simple version, then enhance
• Test incrementally: Verify each piece works before moving forward

Using AI as a Thinking Partner vs. Just a Tool

AI can be more than just a content generator—it can help you think through problems, explore alternatives, and refine your ideas.

As a thinking partner:

• Brainstorm: "What are different ways I could organize this content?"
• Challenge assumptions: "What's wrong with this approach?"
• Explore alternatives: "What are three different ways to solve this?"
• Clarify thinking: Explaining your idea to AI helps you refine it

As a tool:

• Generate boilerplate code or structure
• Format and organize existing content
• Create variations of something you've already defined
• Handle repetitive tasks efficiently

The best results come from using AI in both modes—thinking together, then executing efficiently.

Knowing When to Accept AI Output vs. When to Push Back

AI is powerful but not infallible. Developing good judgment about when to accept suggestions and when to push back is essential.

Accept AI output when:

• It meets your requirements and quality standards
• It solves the problem effectively
• It offers a better approach than you had in mind
• The technical implementation is sound

Push back when:

• It doesn't match your vision or goals
• The technical approach has flaws you can see
• It's too generic and lacks specific context
• It contradicts your expertise in the domain
• Something feels "off" even if you can't articulate why yet

The "Session Diary" Approach

One of the most effective techniques for complex projects is maintaining a session diary—a running record of what you've accomplished and decided in each working session.

Why it works:

• Context persistence: Overcomes AI's lack of long-term memory
• Project continuity: Pick up where you left off without repeating yourself
• Decision tracking: Remember why you made certain choices
• Progress visibility: See how far you've come
• Transparency: Document the collaboration process itself

How to implement:

1. At the end of each session, create a summary of what you accomplished
2. Note key decisions, approaches, and any issues encountered
3. Record what's planned for next time
4. Start the next session by uploading the previous diary entry
5. The AI reads the context and continues seamlessly

Managing the Relationship

What AI is Good At vs. What You Should Do Yourself

Understanding the strengths and limitations of AI helps you delegate effectively and maintain control of your project.

AI excels at:

• Pattern recognition: Following established formats and structures
• Generating variations: Creating multiple versions or examples
• Boilerplate creation: Standard structures, templates, frameworks
• Organizing information: Structuring content logically
• Technical formatting: HTML, CSS, code syntax, markdown
• Explanation: Breaking down complex concepts
• Rapid iteration: Quickly trying different approaches

You should handle:

• Vision and direction: What you're building and why
• Domain expertise: Deep knowledge AI doesn't have
• Quality judgment: Does this meet your standards?
• Creative decisions: What approach feels right?
• Final verification: Testing, validation, accuracy checking
• Unique insights: Your personal experiences and perspectives
• Audience connection: What will resonate with your readers?

Maintaining Your Voice and Vision

When AI helps create content, it's critical to ensure the result sounds like you and reflects your perspective, not generic AI output.

How to maintain your voice:

• Edit actively: Don't just accept AI output—revise it to match your style
• Provide examples: Show AI how you write or explain things
• Give feedback on tone: "Too formal," "too casual," "more technical"
• Insert your experiences: Add personal anecdotes and real examples
• Question generic statements: Push back on AI-isms and clichés

How to maintain your vision:

• Be clear about goals: What are you trying to achieve?
• Set boundaries: "This is a teaching site, not marketing material"
• Make the big decisions: Structure, approach, scope, audience
• Course-correct: If AI takes things in a wrong direction, redirect

Fact-Checking and Verification

AI can be confidently wrong. Always verify important information, especially technical details, statistics, and claims about current events.

What to verify:

• Technical accuracy: Does this code work? Is this command correct?
• Current information: AI's training data has a cutoff date
• Statistical claims: Numbers and percentages need sources
• Specialized knowledge: Domain-specific information in your field
• Best practices: Verify recommendations against authoritative sources

Verification strategies:

• Test code and commands before publishing
• Cross-reference technical claims with documentation
• Use web search for current information
• Apply your domain expertise to spot errors
• Ask AI for sources, but verify them independently

Copyright and Attribution Considerations

When working with AI, understanding intellectual property and proper attribution is important for ethical and legal reasons.

Key principles:

• AI output isn't copyrighted: But you can't claim copyright on pure AI-generated content
• Human creativity matters: Your edits, organization, and original additions create copyrightable work
• Be transparent: If appropriate, acknowledge AI assistance
• Respect sources: Don't ask AI to reproduce copyrighted material
• Create original value: Use AI to help build something new, not to copy existing work

Attribution approaches:

• Full transparency: "Created in collaboration with AI" (like this site)
• Tool mention: "Built with AI assistance"
• No mention: Appropriate when AI was just a tool, like spell-check
• Context-dependent: Academic work may require disclosure; casual writing may not

Putting It All Together

Effective AI collaboration combines all these skills into a working methodology. You bring expertise, vision, and judgment. AI brings capability, speed, and tireless iteration. Together, you can accomplish work that would be impossible—or at least much slower—alone.

The collaboration workflow:

1. Define what you want: Clear goals and requirements
2. Provide context: Give AI the background it needs
3. Start simple: Build incrementally, not all at once
4. Iterate actively: Review, refine, improve
5. Maintain control: You make the final decisions
6. Verify results: Test and validate before trusting
7. Document progress: Track what works and what you've learned