How to Create Effective AI Prompts that Generate Tangible Results

Find your rhythm: create effective AI prompts

AI-powered tools are coming closer to becoming standard in the workplace—some 27 percent of employees use AI in their daily workflow, nearly doubling usage rates in just two years. As is typical with all new technologies, there are innovators, early adopters, laggards, and everyone in between.

For those innovators and early adopters, writing AI prompts is a cinch—like riding a bike or sending an email—but for those of us in the middle, it’s not so easy. And for good reason: working with and writing AI prompts that produce effective results is a learned skill.

Prompting an artificial intelligence can be like getting to know a stranger—one that doesn’t speak unless spoken to, putting the obligation to engage on you, the user. The exception, of course, being AI chatbots on retail websites—the digital descendants of Microsoft’s Clippy. Either way, it might take several conversations to find a good rhythm.

In this post, we’ll cover exactly what an AI prompt is, strategies for successful prompting, the pros and cons of AI writing, and examples of effective AI prompts on a scale from basic to complex.

What is an AI prompt?

An AI prompt is what you input into an AI system to obtain certain information. It can be a simple question or phrase, a complex sentence, or even full paragraphs of input. The detail, specificity, tone, and even accuracy of the response depends on the wording of your prompt.

The advanced version? Writing AI prompts is like starting a conversation with a machine that you’re programming in real time. You start with a simple question—and receive a simple answer—but you can build a continuing prompt. The AI will respond accordingly, taking into account what’s already been said.

12 strategies for successful prompting

The quality of AI responses—or output—is directly proportional to your ability to write effective prompts. You may not get what you’re looking for on the first couple of tries, but with practice, you can be a pro at writing AI prompts in no time.

Here are 12 quick strategies for writing AI prompts successfully.

1. Be specific

The more detailed your prompt, the better response you’ll get. Consider the difference between prompting a model with “Write me a story” and “Write me a 1,500 word science fiction story about a robot contemplating what it means to be human.”

2. Define the format and style

Clearly state what format you want the response to be in. Are you looking for an outline? A bulleted list? A poem in the style of Emily Dickinson or Sylvia Plath?

3. Set the tone

Indicate the desired voice you’d like the response to mimic—professional, casual, academic, persuasive, or even, “Write an email to my boss explaining why I need a vacation in the heroic tone of Homer’s The Odyssey.”

4. Identify the audience

Tailor the prompt for who the response is intended for. This could be high school students, marketing professionals, or crime drama fans.

5. Give context

Be sure to provide any necessary background information to help AI understand the context of your prompt.

6. Limit the scope

Focus the prompt to avoid overly broad results. Consider the difference between “List the benefits of workplace training” and “List the benefits of workplace training as it pertains to reducing customer call center average handling time.”

7. Use constraints, if needed

Consider adding word count, formatting, or structure limits to avoid lengthy, vague results. Prompting AI for “The challenges of employee training” may return a long generalized list of challenges. Try “List, in bulleted format, six of the most common challenges faced by learning and development programs when developing onboarding for new sales reps.”

8. Ask follow-up questions

The initial response may include complex wording or industry jargon you’re unfamiliar with. Prompt AI to expand, clarify, and refine output as needed.

9. Avoid ambiguity

Use clear, concise language to reduce irrelevant responses. Consider the difference between “Explain basketball to me” and “Explain how the rules of basketball have changed from the 1970s to the modern era, specifically surrounding the addition of the three-point line and its effects on average points scored per game.”

10. Experiment and iterate

Notice you’re not getting the results you’re after? Play around, experiment, and try different wording. You’re not just training the AI to respond effectively to your inquiries, you’re training yourself to write better AI prompts.

11. Use examples

You may need to provide a reference or example if your prompts aren’t producing the output you’re looking for. Provide a short sample of a specific writing style for the AI to examine as part of your prompt.

12. Indicate creativity level

Without instruction, AI’s output will likely sound clinical and bland, which might be what you need. Clarify whether you’re looking for factual, imaginative, or a mix of factual and creative content.

While these AI prompt writing strategies cover how to be specific and provide enough detail for quality responses, sometimes it’s the type of prompt that matters.

Types of AI prompts

As you progress in your journey, it’s helpful to understand that more and more detail isn’t the only way to approach writing AI prompts more effectively. Read on to discover six common types of AI prompts, presented in order from basic to complex.

Zero-shot prompt

A zero-shot prompt is the most simple type of prompt in which the user gives basic, clear instructions without providing any examples or references. These types of prompts are useful for quick, general responses, most often used for translation, question answering, text summarization, or basic text generation. Let’s use basic text generation as an example.

Zero-shot prompt example: “Write a short poem about the fleeting nature of life.”

While this prompt does define the format (poetry), provides vague constraints (short), and a subject to write about, it doesn’t provide an example or reference to indicate tone, style, or audience.

Few-shot prompt

Going from zero to a few, you might be able to guess where this one is headed. A few-shot prompt provides a few examples of what you want the AI’s response to imitate. This helps the model understand the tone and structure you’re looking for. Let’s continue using the previous example as a comparison.

Few-shot prompt example: “Here are three poems written by Mark Strand [copy/paste poems here]. Write a poem about the fleeting nature of life in the same style.”

In this example, after the initial prompt, you would copy, paste, and send the reference material. You can imagine how much more accurate the AI’s response will be when compared to the zero-shot prompt we started with. By providing actual writing samples, the AI will now be trained on the style, tone, and structure you desire.

Instructional prompt

Instructional prompts begin with direct commands like “write,” “summarize,” “list,” “explain,” or “compare.” Essentially, starting your prompt with an instructional command elevates it from the most basic category of zero-shot prompt—question answering—into more advanced AI prompting.

Our poetry example already begins with the direct command “write,” but we can modify it in several ways, adding constraints as we go.

Instructional prompt examples:

• “Summarize this poem by Mark Strand.”
• “Explain the meaning of this poem by Ada Limón for a high school audience.”
• “Compare the use of metaphor in this poem by Michael Bazzett [insert poem] to that of this poem by Dan Beachy-Quick [insert poem]. Write in a persuasive tone for an academic audience familiar with their work, and keep your analysis under 500 words.”

As you can imagine, adding constraints, providing examples, and indicating desired tone and audience will help to filter the AI model’s response to your expectations. The more you experiment and iterate, the more possibilities you’ll see.

Role-based prompt

Having fun yet? If not, role-based prompts might be just what you need. Role-based prompts ask the AI model to assume the identity—tone of voice, values, beliefs, experiences—of a known persona or viewpoint. This could be a general persona like a criminal defense lawyer or an actual person like Thomas Jefferson.

Role-based prompt examples:

• “You are a criminal defense lawyer on the popular cable-TV show CSI: Miami. Write a 500-word opening statement defending the Sesame Street character Cookie Monster, who is facing a first-degree arson charge brought against him by his favorite bakery.”
• “You are Thomas Jefferson. Write a diary entry expressing a mixture of determination and hesitancy following the signing of the Declaration of Independence. Include minor details like what you ate that day for breakfast and your annoyance due to an itchy powdered wig.”

As you can see, role-based prompting allows for limitless creativity, but it’s good for practical purposes, too. Professionals of all types can prompt AI to write in the style of their own role, which we’ll cover in the last section.

Contextual prompt

Contextual prompts begin with relevant context or background information that can potentially change the AI output. This is especially useful when tailoring responses within certain constraints or editing reference material.

Contextual prompt examples: 

• “This is for a 9th grade modern American literature course [insert content]. Rephrase it in simpler language.”
• “This will be presented at an annual meeting of financial advisors [insert content]. Adjust the tone to be more formal.”
• “This is a best-man speech for my best friend’s wedding [insert content]. Rephrase it assuming there are innocent children in the audience.”

The context given at the beginning of each prompt dictates the style, tone, audience, or setting that the model’s response follows, giving you more control over the results.

Meta prompt

Meta prompts are often reserved for professionals training the AI models before they reach the public, but they’re still useful if needed. They act as system-level instructions that inform the AI’s behavior, tone, and scope.

Meta prompt examples:

• “Always cite real sources. Never guess.”
• “Always respond in a formal tone.”
• “Adhere to APA Style citation rules.”

Pros and cons of AI-generated writing prompts

As with any newly adopted technology, there are both advantages and limitations. Let’s take a look at both, and finish with some tips on how to find a good balance.

Advantages

The right AI tools can jumpstart your writing if you’re drained of energy. If you have no one to bounce your ideas off of—the case for many deskless and remote workers—AI can be your brainstorming buddy. In addition, AI-generated writing prompts can help:

• Organize messy first drafts
• Improve the clarity, style, or tone of your writing
• Rephrase content for specific audiences
• Summarize complex ideas
• Improve SEO metrics

In general, AI writing prompts can help boost productivity and be a time-saving tool when facing looming deadlines.

Limitations

Where AI writing truly falls short is matching human creativity, and that’s okay. AI tools and models don’t have real-life experiences and memories to draw from. They can imitate it, but when it comes to search engine optimization (SEO) and search rankings favor human writing. Other limitations include:

• Potential inaccuracies and plagiarism
• Reproduction of harmful or unwanted bias
• Data privacy risks

Human + AI collaboration tips

To successfully integrate AI into your daily workflow, take into consideration the following tips:

• Stay in the driver’s seat. Don’t let AI take over your entire writing process. If relied on too much, you’ll start to lose your own sense of creativity, identity, and voice—everything that makes up the irreplaceable human touch.
• Fact-check. Some AI models are prone to inaccuracies. Be sure to cross-check with trusted sources.
• Use it as inspiration. As mentioned in the advantages section, use it as a brainstorming buddy, a last resort when writer’s block has set in, or as a general sounding board for new ideas.

AI prompts and e-learning

As a busy learning and development professional, being able to write effective AI prompts can be a massive help. The increased productivity and efficiency allows you to generate course summaries in seconds, transform source material into relevant content, write engaging quiz questions, and refine course content for specific learner groups at the click of a button.

Ready to test your new AI prompt-writing skills? Start your free trial today of the Articulate 360 platform, and get access to a powerful AI co-pilot that can help you build better courses up to 9x faster.