Using ChatGPT Effectively: Common Mistakes and Solutions

Introduction

• Self-proclaimed experts often provide advice on new features like ChatGPT without sufficient testing.
• Example: Custom instructions should not be overly specific.

Custom Instructions

• Best Practices:

  • Keep it simple: 3 sentences.
    1. Professional snapshot (e.g., job role).
    2. Secondary persona (interests outside of work).
    3. Personality traits and interests.
  • Example for instructions:
    • I am a product marketing manager in tech.
    • I create content on various topics.
    • I value efficiency and productivity.

• Second Custom Instruction Box:

  • Use a respected creator as a model (e.g., Ali Abdaal).
  • Analyze their writing style through their content (e.g., blog posts) and create instructions for ChatGPT.
  • Pick styles that resonate with you (e.g., casual language, sarcasm).

Automating Tasks with Code

• Many users avoid writing code to automate tasks.
• Example: Creating a Google App Script for meeting notes:
  • Request functionality: add date, notes, and action items.
  • ChatGPT can generate code and provide implementation steps.
  • Easy tweaks can be made to the code for personalization.

The 'First Try' Fallacy

• Users expect perfect responses from the first prompt.
• Solution: Add a follow-up request at the end of prompts:
  • Ask ChatGPT to pose 5 questions to clarify your request.
  • Example: Seeking help with tracking currency payments in Google Sheets.
  • This leads to more relevant and actionable responses.

Summary Only Shortfall

• Users often summarize information instead of seeking actionable insights.

• How to Improve:

  • Provide context when asking for summaries (specific to your role).
  • Example: "How does this article apply to me as a paid media marketing manager?"
  • Can upload PDFs for detailed analysis.

• Q&A Sessions:

  • Instruct ChatGPT to assume a specific role for targeted Q&A.
  • This allows for focused analysis based on provided content (e.g., reports).

The Prompt Overload Paradox

• Users collect prompts but fail to implement them.
• Steps to Address This:
  • Instruct ChatGPT to act as a prompt engineer and request productivity prompts tailored to your job.
  • Choose 3 prompts to implement consistently for two weeks.
  • Refine these prompts using feedback and experiences.

Conclusion

• Focus on practical implementation of ChatGPT to enhance productivity.
• Avoid getting overwhelmed with excess information; refine and apply effectively.