Prompt Engineering - Make The Most of GenAI

What is Prompt Engineering?

Prompt engineering is the art and science of designing input queries (or prompts) to maximise the performance of Generative AI models. Instead of programming the AI to solve a problem explicitly, you create well-structured prompts that allow the model to generate the desired response. It's like using natural language to code.

For instance, asking a GenAI model "Can you explain what machine learning is in simple terms?" is a prompt. By adjusting how you frame questions, provide context, or even add constraints, you can achieve more relevant, detailed, or creative responses from the model. We're going to discuss some hints and tips below, but first...

Why is Prompt Engineering Important?

Generative AI models are flexible but also sensitive to how you ask questions. A poorly designed prompt might lead to irrelevant, vague, or biased responses. Getting better at prompt engineering allows you to:

• Generate higher-quality outputs by giving the AI clear instructions.
• Solve problems more efficiently by guiding the model to the desired results.
• Increase creativity by experimenting with different ways of structuring prompts.

Core Concepts in Prompt Engineering

Clarity and Specificity

A good prompt clearly conveys your intent. The more specific the prompt, the more accurate and tailored the response. Avoid ambiguous language and unnecessary complexity.

Example:

• Poor prompt: "Tell me about cars."
• Improved prompt: "Give me a brief history of electric cars and how they differ from traditional gasoline cars."

The second prompt provides the model with more context and direction.

Role-Playing Prompts

Instructing the model to take on a specific role can help generate responses that match a desired style or tone. This works well when you need specific types of outputs, such as an explanation from the perspective of an expert, a child, or a historical figure.

Example:

• Standard: "Explain the theory of relativity."
• Role-playing: "You are a science teacher explaining the theory of relativity to a 12-year-old. Keep it simple and use easy-to-understand examples."

This role-based approach helps tailor the response to the expected audience or style.

Context and Constraints

Providing context and constraints can help the AI understand the boundaries and requirements for its response. Constraints like length limits, tone (formal/informal), or specific focus (like technology vs. history) will result in more tailored outputs.

Example:

• Without constraints: "Write about artificial intelligence."
• With constraints: "Write a 150-word summary on how AI is used in healthcare, in a conversational tone."

The second prompt includes context (healthcare) and constraints (length and tone), guiding the AI to generate a more focused response.

Step-by-Step Instruction

For complex or multi-step tasks, breaking down instructions into smaller steps can help the AI follow a logical flow. This technique is especially useful when asking the AI to perform tasks like writing code, solving maths problems, or summarising long articles.

Example:

• "Summarise this article and list the key points."
• Improved: "First, summarise the main idea of the article in one sentence. Then, list the three key points covered in the article."

The second approach uses step-by-step instructions, which helps generate a more organised output.

Providing Examples

When the task is creative or open-ended, providing examples in the prompt can help the AI understand the style, format, or structure you expect.

Example:

• "Write a poem about the ocean."
• With an example: "Write a four-line poem about the ocean, similar to this style: 'The waves crash high upon the shore, / A dance of nature, evermore.'"

Providing examples helps narrow down the creative direction of the model.

Tomorrow, we'll look at specific techniques that you can deploy to test and hone your prompt engineering talents.

--

?? Hi, I’m Jenny.? I’m VP of AI Innovation at Oracle’s AI Office.? Before my current role? I ran an AI company for over a decade, so I’ve witnessed first hand the impact AI can have on companies, both large and small.? My team and I are on a mission to make everyone an AI hero, which is why you can follow me for free here on LinkedIn to become an expert.? Please subscribe and share to help spread the knowledge.? I'm creating this newsletter with the help of my robot buddies.

?