glitter pile ai

Today I completed the first step of my personalized study plan: The Ultimate ChatGPT Technical Self-Study Bootcamp, which includes 30 videos. The concepts are really familiar to me already because I spent most of May this year studying prompt engineering on Coursera, specifically the Vanderbilt University certification program taught by Jules White. What I enjoyed about the new series was getting refreshed again on all of the same concepts but from a new perspective and with very different use cases. For example, Tina used prompts related to learning language, anime (Naruto!) and study plans, whereas Jules focused primarily on uses related to academic tools, like breaking down campus policies and creating content for professors. Both have been super helpful!

Prompt Patterns Covered (and my definition of each):

• Zero Shot
  • Asking questions without any format or examples
• Zero Shot Chain of Thought (“CoT”)
  • Asking questions without examples but asks the gen AI to “let’s think step by step” which encourages the model to generate a chain of thought that answers the question.
    • Alternative Approach: Plan and Solve (PS)
      • Planning: First, the model devises a plan to dived the entire task into smaller subtasks.
      • Execution: Then, the model carries out the subtasks into smaller subtasks.
• Few Shot
  • Provide a few examples as an outline/framework for the intended output organization
• Cognitive Verifier
  • Asking gen AI to explain how it arrived at the output
• Persona, and inversely, Audience
  • AI gives output in the identify/personality of a person, or inversely, provides a reaction or content to you as a selected personality
• Flipped Interaction
  • Get AI to interview you/ask you questions in order to shape the answer more granularly, especially helpful if you don’t quite know where to start
• Chain of Thought
  • Formatting the desired output by providing examples of how the output should be organized
• Directional Stimulus
  • Hinting to AI for the result
• Gameplay
  • Make it fun!
• Template
  • Keep output organized to your preference)
• Recipe
  • When you only have the start content and goals to finish but need the middle/help thinking of more steps)
• Alternatives
  • Creating more options outside of your original outline including different approaches so you have options, and can compare and contrast solutions
• Outline Expansion
  • Creates outline for complex problems to help break it down into more manageable sections iteratively
• Fact Checks
  • Requiring gen Ai to show where it got the content, especially important when you are not an expert on the topic and unable to verify the accuracy of the content yourself
• Tips and Tricks
  • use gen Ai to create more prompts for you
  • manage length of content
  • provide direction to manage length of response
  • using outlines