0% found this document useful (0 votes)
16 views23 pages

Prompt Engineering Farmer Presentation

Uploaded by

nsadat1925
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
16 views23 pages

Prompt Engineering Farmer Presentation

Uploaded by

nsadat1925
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 23

PROMPT

ENGINEERING 101:
ASKING GOOD QUESTIONS
WHAT’S THE PLAN?

• Master the fundamentals of effective prompt design.


• Learn to apply prompts in diverse scenarios, from academic
research to creative content generation.
• Gain hands-on experience with prompt refinement and
iteration.
WHAT IS PROMPT ENGINEERING?

• The process of crafting precise, effective questions or


instructions to achieve specific, high-quality responses from
AI systems.
• Precision in prompts:
• Saves time
• Reduces misunderstandings
• Improves efficiency.

• The quality of AI output directly relies on the clarity and


relevance of the prompts it receives.
WHY THE “RIGHT” QUESTIONS MATTER

• Ensuring Clarity
• AI tools interpret inputs literally.
• Ambiguous or unclear prompts lead to generalized, less effective responses.
• Clear questions foster direct, helpful answers and make follow-up prompts easier to construct.

• Promoting Relevance and Specificity


• Targeted prompts narrow down the focus, helping AI skip irrelevant information and hone in on what
matters.
• “Summarize three major economic impacts of climate change” is more effective than “Tell me about climate
change.”

• Enhancing Efficiency
• Thoughtful prompts maximize the efficiency of AI tools, saving time and resources.
• Effective prompting is an iterative process that involves refining questions to be as direct as possible.
• A repeated process of making gradual improvements
KNOWING ITS LIMITS AND ITS STRENGTHS

• Strengths
• Pattern Recognition
• AI is adept at processing vast data sets and identifying patterns, such as spotting trends in literature, summarizing large texts, or
generating visual ideas from text.

• Content Generation
• Can produce new content in various formats—text, images, and code—based on learned data.

• Limitations
• Context Understanding
• AI can lack human nuance and understanding of complex contexts.

• Creativity Constraints
• AI follows given instructions and rarely "thinks outside the box" unless explicitly instructed.

• Implications for Prompt Engineering


• To maximize AI's potential, prompts should be designed to guide the AI in ways that account for its strengths while navigating around
its limitations.
TYPES OF PROMPTS

• Open Prompts
• Encourage broad exploration (e.g., “Explain how AI is changing education”).
• Suitable for brainstorming or idea generation.

• Closed Prompts
• Aim for specific answers
• Useful for focused, concise responses.
• “List three ways AI impacts education.”

• Descriptive Prompts
• Require the AI to describe or explain
• “Describe the theme of identity in To Kill a Mockingbird.”

• Directive Prompts
• Ask AI to perform an action, like summarizing, comparing, or generating a list.
• “List three ways AI impacts education.”

• Exploratory
• Broaden the conversation, seeking multiple perspectives
• “What are various theories on human intelligence?”

• Conclusive
• Aim for specific, clear conclusions or answers
• “Summarize the main argument of the article.”
COMPONENTS OF AN EFFECTIVE PROMPT

• Context
• Provide necessary background (e.g., “Based on recent developments in machine learning…”).
• The context helps guide the AI to provide responses that are accurate and relevant.

• Task Clarity
• Specify what the AI should do. Is it summarizing, analyzing, listing, or comparing?
• Instead of “Explain Shakespeare,” specify “Summarize the themes in Hamlet with examples.”

• Constraints
• Set clear parameters, such as word limits, tone, or style.
• “Summarize in 50 words or less” or “Provide a response in an academic tone.”

• “Describe the historical impact of the Renaissance on art and literature in no more than
100 words, focusing on key artists and writers.”
PROMPTING GUIDELINES

• Use Clear Language


• Avoid jargon, acronyms, and ambiguous terms. Use simple, direct language.
• Replace “Discuss the ontology of AI” with “Explain what AI is and how it works.”
• Be Specific
• Vague prompts lead to general answers.
• Define exactly what you need.
• Instead of “Tell me about AI,” try “Explain three key benefits of using AI in
healthcare.”
• Define the Outcome
• Specify what you want: a summary, list, comparison, argument, etc.
• “List three reasons why AI is beneficial for small businesses and briefly explain
each.”
COMMON PITFALLS

• Overly Broad Prompts


• Effective prompts communicate SPECIFICALLY what task the AI is to perform
• Rather than asking an AI agent to “Explain history,” you might request that it
“Summarize the impact of WWII on the American economy.”

• Avoiding Assumptions
• Don’t assume AI knows specific contexts;
• Provide clear context about what you are attempting to answer or create
• Rather than asking, “Describe this author’s impact,” specify the author’s name and
field.
KEY CRITERIA FOR EVALUATING PROMPT EFFECTIVENESS

• Clarity
• Is the prompt easily understandable?

• Relevance
• Does it directly address the desired outcome?

• Specificity
• Is it detailed enough for a targeted answer?

• Alignment
• Does it align with the intended goal?
EFFECTIVE OR INEFFECTIVE?

• In a 100-word summary, explain why chickens should be considered valuable


employees in a startup focused on organic pest control.
• Tell me about pineapple pizza.
• List three scientific reasons why a sandwich with extra pickles could
potentially improve someone’s mood. Explain each reason briefly.
• Write about a robot dog.
• Create a 200-word lesson on the physics of spaghetti noodles in space,
explaining how astronauts could cook pasta without gravity.
• What’s the deal with noodles?
• List three reasons a squirrel might be the best choice as a forest tour
guide. Make each reason humorous but believable.
PROMPT MAKEOVER

• You have been given a simple and flawed prompt


• Using the skills that we have learned today, improve the prompt
• Give it a makeover!
• Give it a glowup!
GLOW UP EXAMPLES

• Original Prompt: “Tell me about exercise.”


• How might we improve this prompt?
• “Describe the benefits of running as a form of exercise, focusing
on cardiovascular health and mood improvement, in 75 words or
less.”
• Original Prompt: “Explain what animals do.”
• “List three unique survival strategies animals use in the wild,
with examples, in 100 words.”
ADVANCED TECHNIQUES

• Chain of Density Prompting


• Step-by-step approach where AI follows a logical chain of reasoning.
• Example: “Describe the pros and cons of social media, then evaluate its
impact on mental health.”
• Essay Compression Prompting
• This technique is useful for condensing longer, more detailed information
into a shorter, easy-to-read summary.
• Role-Playing Prompts
• Using persona-based prompts like “As an economist, analyze…” to improve
contextual relevance.
CHAIN OF DENSITY PROMPTING

• Guide the AI to produce responses that are:


• Dense with relevant information
• Structured, and in-depth, making it easier to achieve high-quality outputs in
complex topics.

• Why COD Prompting Works:


• Each layer (or “density step”) helps the AI focus on specific aspects,
preventing overly broad responses and encouraging well-rounded answers.
• The final prompt includes multiple layers that guide the AI to provide a
detailed, organized response, using examples, comparisons, and structured
content.
HOW-TO COD PROMPT
• Start with a Basic Question
• Begin with a simple question or task, keeping it broad and easy to understand.
• “How does a penguin becoming mayor help the town?”

• Add Detail in Small Steps


• Take the fundamental question and add more information to make it specific.
• This is like “zooming in” on one part of the topic.
• Explain how a penguin mayor helps reduce littering and keeps the streets icy (in a good way).

• Break the Question into Parts


• Add another step to focus on specific areas or comparisons, so the AI knows to give a complete answer.
• Compare how a penguin mayor’s approach to town cleanup is different from a human mayor’s approach.

• Ask for Examples to Make It Real


• Including examples helps the AI give a more concrete answer that feels real and relatable.
• Give one funny example of a penguin mayor cleaning up the town or getting penguins to recycle fish wrappers.

• Organize the Answer by Sections


• Request the answer in parts or sections to keep it well-organized and easy to read.
• “Summarize in two sections: one about how a penguin mayor improves the town, and one comparing their methods to a
human mayor’s.
FINISHED COD PROMPT

• Explain how a penguin mayor helps reduce littering and keeps the
streets icy (in a good way). Compare these benefits to what a
human mayor would do, and give a funny example of a penguin mayor
helping penguins recycle fish wrappers. Summarize in two sections:
one about the penguin mayor’s improvements and one comparing them
to a human mayor.
CHAIN OF DENSITY RACE

• The next slide contains simple prompts that need to be expanded using
the Chain of Density prompting method
• A prompt will be revealed then we will take two minutes to add ONE
DETAIL OR ONE CONSTRAINT to make the prompt more effective
• We will repeat this process three times.
• After each round, we will share the prompts that we created.
• The prompt that the group selects will be the new prompt that is to be
expanded or constrained
C.O.D. RACING

1.“Explain how a library works.”


2.“Describe the importance of sleep.”
3.“Talk about the benefits of cooking at home.”
4.“Explain what makes a story interesting.”
5.“Describe what a good friend is.”
ESSAY-COMPRESSION PROMPTING

• Useful for condensing longer, complex information into clear, focused


summaries, ideal for generating concise responses.
• Rather than asking,“Summarize this text,” this technique preserves more of
the original language of the provided text, ensuring great clarity,
preciseness, and specificity

• Setting a word limit, key points, and a clear structure makes it


simpler for the AI to produce a focused summary without overwhelming
information.
HOW TO ESSAY-COMPRESS
• Start with the Big Question or Topic
• Begin by stating the main idea or question you want summarized.
• Think of it as the “big picture” that you want to compress.
• “Summarize the benefits of hiring a llama as an office manager.”

• Set a Clear Word or Sentence Limit


• Specify a length for the response to keep it concise and focused.
• “Summarize in 30 words or less.”

• Identify Key Areas or Points to Cover


• Pick 2-3 main points or ideas that you want included.
• This helps the AI focus on the most important aspects of the topic.
• “Include benefits like boosting team morale, handling paperwork with ease, and being eco-friendly.”

• Add Structure to Guide the Summary


• Ask for an organized answer, like a bullet-point list or short paragraphs, to make the summary easy to read.
• “Summarize in 2-3 bullet points.”

• Ask for Clear Language


• Specify that you want a simple or beginner-friendly answer, especially for complex topics.
• “Explain in a fun and simple way that anyone could understand.”
SUCCESSFUL ESSAY-COMPRESSION PROMPT

• Summarize the benefits of hiring a llama as an office manager in


30 words or less. Include benefits like boosting team morale,
handling paperwork, and being eco-friendly. Use 2-3 bullet
points, in simple, fun language.
WRAP UP

• Key Takeaways
• Emphasize the importance of clarity, specificity, and
purposeful prompting in generating effective AI responses.
• Prompt engineering is a skill that grows with practice and
iterative refinement.
• Final Thought
• Encourage attendees to experiment with prompts, using the
principles learned today to maximize the usefulness of AI tools
in their personal and professional lives.

You might also like