Prompt engineering is the process of carefully designing the input (or “prompt”) you give to an artificial intelligence (AI) system, like ChatGPT, to get the best possible response or outcome. Think of it like giving clear and specific instructions to a very smart helper.
For example, if you want the AI to explain something, solve a problem, or create a story, the way you phrase your request makes a big difference in the quality of the answer. It’s about figuring out how to ask the question or give the task in a way that the AI understands what you’re asking for.
Good prompt engineering often involves being specific, using examples, or setting the right context for the AI to work with. It’s like being a great teacher or guide for the AI, helping it understand exactly what you need.
Your first question may be.... "what the heck is a delimiter?!" As you will see below, they are symbols that you can add to your prompts to get a desired response. Below are examples of each and how you can use them to achieve a better output!
See an example chat of each of these at work!
*Note: The chat was made with the idea of making a separate "Cavas" for each delimiter. This would provide a more concise explanation and room to explore.. As of 1/4/25, you cannot see those in the shared chat. Feel free to make your own chats with these examples!
Specific Purpose: Used when you want to emphasize exact words, phrases, or quoted text. It tells AI, “This is the precise text you should focus on.”
Example:
Exact Word or Phrase: Explain what “photosynthesis” means.
Here, you’re asking AI to specifically define the term as written.
Direct Quote or Text: What does it mean when someone says, “Knowledge is power”?
This indicates you’re asking about the meaning of the quoted phrase as a whole.
Specific Purpose: Brackets are for additional context, clarification, or instructions about how to tailor the response. They modify or refine the primary request.
Examples:
Clarification: Explain the rock cycle [in simple terms].
Here, the brackets clarify that the response should be simplified.
Specific Audience or Scope: Describe ocean currents [for 10th-grade students].
This instructs AI to adjust the explanation for a specific audience.
Specific Purpose: Parentheses are for optional or supplementary information. The main idea is still clear without the parentheses, but the extra details add depth or provide alternatives.
Examples:
Additional Information: Define plate tectonics (include examples of convergent and divergent boundaries).
The parentheses add an optional element to expand the response.
Alternatives: How do tides work (spring or neap tides)?
This suggests the explanation can focus on either type of tide, providing choice.
What is the difference between brackets and parentheses?
The brackets are used as instructions, or to provide scope and context. Brackets will change how the question is answered by AI.
Parentheses are for supplementary materials and information you would like to include. It adds extra context without changing the core question.
Purpose: Often used for grouping related elements, defining key-value pairs, or indicating optional/repeatable sections.
Examples:
Key-Value Pair: {“term”: “photosynthesis”, “level”: “beginner”}
Indicates structured data, such as asking for a beginner-level explanation of photosynthesis.
Grouping/Repeating Elements: Provide examples of renewable energy sources {solar, wind, geothermal}.
Curly braces group related terms.
*How are these different from parentheses: The curly brackets provide a portion of the core answer. So, in parentheses, you may not see those specific examples, if suggests one. But in curly brackets, you are telling AI to specifically include those within a group.
Specific Purpose: XML-style tags allow you to structure your request explicitly. Tags can define parts of the input (like categories, formatting, or focus areas) or specify output formats. They structure complex requests clearly, especially when asking for multiple items or specific formats.
Examples:
Tagging a Request:
<topic>Earth Science</topic> <focus>volcanoes</focus> <level>high school</level>
This tells AI you want information about volcanoes for a high school level Earth Science lesson.
Output Formatting:
<question>What causes earthquakes?</question> <format>bullet points</format>
This indicates you want the answer in bullet-point format.
Multi-Part Question:
<step1>Explain latitude and longitude.</step1> <step2>Describe how to find the altitude of Polaris.</step2>
This organizes complex or multi-step prompts.
Purpose: Separates distinct options, often with the intent of mutually exclusive choices.
Examples:
What is the state of matter: solid | liquid | gas?
Asks to choose between one of these options.
Explain the greenhouse effect | ozone depletion.
Distinguishes two completely different topics for focus.
Purpose: Adds comments or meta-instructions to the prompt.
Examples:
Explain density // Use simple terms for a middle school audience.
Indicates the main question but adds a comment to tailor the response.
Create a diagram // Focus on convection currents in the mantle.
Asks for a specific type of output but includes additional guidance.
Purpose: Used when you want to highlight or call out a specific piece of text. Tells AI, “This is a quoted or emphasized section, distinct from the rest.”
Examples:
> Use this quote to guide your explanation:
> "Education is not the filling of a pail, but the lighting of a fire."