Lesson 7: Effective prompting techniques (Deep Dive) | AI Fluency: Framework & Foundations Course

Key Concepts

• Prompting: Communicating clearly with AI to achieve desired outcomes.
• Prompt Engineering: Designing effective instructions for AI systems.
• Context: Providing background information to guide the AI's response.
• Few-shot/N-shot Prompting: Providing examples for the AI to emulate.
• Output Constraints: Specifying desired format, length, or style of the AI's output.
• Chain of Thought Prompting: Breaking down complex tasks into smaller steps for the AI to follow.
• Role Definition: Specifying the persona or expertise the AI should adopt.
• Iterative Prompting: Refining prompts based on the AI's responses.
• Artifacts: Unique Claude outputs that may be easier to understand or more interesting to digest.

1. The Essence of Prompting

Prompting is presented as a practical skill, akin to explaining a task to a new colleague. It emphasizes clear communication of what is wanted, how it should be done, and how to interact with the AI assistant. The video uses Claude as the primary example, but the principles are applicable to other AI systems. Prompt engineering is defined as the practice of designing effective instructions for AI systems. Effective prompting combines human communication skills with AI-specific considerations.

2. Foundational Prompting Tips

The video outlines six foundational prompting tips:

2.1. Give Claude Context

• Specificity and Clarity: Be specific about what you want, why you want it, and who you are.
• Example: The prompt "Tell me about climate change" is improved to "Explain three major impacts of climate change on agriculture in tropical regions with examples from the past decade."
• Adding Purpose: Further enhanced to "Explain three major impacts of climate change on agriculture in tropical regions with examples from the past decade. I'm preparing for a job interview at an agricultural research lab in Indonesia... write a summary of key concepts that would help me speak intelligently in the interview."
• Benefit: Tailors the response to the specific situation and knowledge level.

2.2. Show Examples of What Good Looks Like (Few-shot Prompting)

• Concept: Providing examples for the AI to emulate.
• Example: Instead of just asking "Please convert this technical statement to plain language: The platform implements end-to-end encryption protocols to safeguard data integrity," the prompt is enhanced with examples.
• Example Prompts:
  • "Original: the quantum algorithm exhibits quadratic speed up. Plain: the new method solves problems roughly twice as fast as previous methods."
  • "Original: the interface leverages intuitive design paradigms. Plain: the design is easy to understand and use."
• Guidance: Cover the full diversity of possible prompts with examples that cover different cases or styles.

2.3. Specify Output Constraints

• Concept: Being clear about desired format, length, language, or other constraints.
• Example: "Create a clean modern single page art portfolio website. Include these main sections: hero about me skills portfolio projects experience and contact. Make the navigation menu sticky and responsive with hamburger menu on mobile. Use a sunset color palette and add a dark light mode toggle in the navigation."
• Benefit: Helps Claude structure its response to match expectations.

2.4. Break Complex Tasks into Steps (Chain of Thought Prompting)

• Concept: Listing out task steps to guide the AI's process.
• Example: Instead of "Analyze this quarterly sales data," the prompt becomes "I'd like to analyze this quarterly sales data. Please approach this by looking through our sales records to identify the top performing products comparing current quarter results to the previous quarter highlighting any unusual trends or patterns and then suggesting possible reasons for these trends."
• Benefit: Ensures Claude follows the desired process.
• Note: Modern reasoning models are increasingly capable of performing step-by-step reasoning on their own.

2.5. Ask Claude to Think First

• Concept: Explicitly giving the AI space to work through its process before executing the task.
• Example: "Before answering please think through this problem carefully. Consider the different factors involved potential constraints and various approaches before recommending the best solution."
• Benefit: Produces more thorough and well-considered responses.
• Note: Modern reasoning or extended thinking models by default think before acting.

2.6. Define Claude's Role, Style, or Tone

• Concept: Specifying how you want Claude to communicate and behave.
• Example: "Please explain how rainbows form from the perspective of an experienced science teacher speaking to a bright 10-year-old who's interested in science."
• Example: "As a UX design expert review this website wireframe and suggest three improvements focusing on user navigation and accessibility."
• Benefit: Guides both Claude's interaction and the final result.

3. Prompt Improvement and Iteration

• Asking Claude for Help: The most powerful technique is asking Claude to improve your prompt.
• Example: "I'm trying to get you Claude to help me with goal I'm not sure how to phrase my request to get the best results Can you help me craft an effective prompt for this?"
• Iterative Process: Effective prompting is iterative and experimental.
• Refinement: If a response isn't quite right, refine the approach by adding specificity, providing examples, breaking down tasks, or trying different techniques.
• Variations: Request different versions or formats of the output.
• Confidence Check: For factual questions, ask "How confident are you about this answer?"
• Conversation Reset: Sometimes starting a fresh conversation gives better results.

4. Guidance and Common Mistakes

• Effective Patterns:
  • Starting with a clear task overview statement.
  • Including format specifications and examples.
  • Setting explicit constraints or requirements.
  • Providing rich and relevant background information.
• Common Mistakes:
  • Assuming that Claude can read your mind.
  • Overloading a single prompt with multiple unrelated tasks.
  • Being too vague about what success looks like.
  • Not providing feedback on previous responses.

5. Synthesis/Conclusion

Effective communication with AI systems like Claude combines human communication principles with AI-specific techniques. The six principles, along with asking Claude for help, form a solid toolkit for AI interactions. Iteration and practice are key to improvement. Prompt engineering is an evolving practice, and while specific techniques may become less necessary as models improve, the principles of good communication remain relevant. Experimentation and adaptation are crucial.