Whether it is by typing a ChatGPT prompt, or engaging in fruitful discussions, language is the key that unleashes the unmatched capabilities of Artificial Intelligence (AI).
In a world where technology dominates, people are becoming more and more worried about the potential of being overridden by AI tools such as ChatGPT. To conquer fear, you need to learn how to use prompts effectively and understand their formulas for delivering different responses.
With the right ChatGPT prompt, you have the power to request a comprehensive business plan or a straightforward recipe for creating zucchini pasta that will please even the most finicky eaters. It can even outline the necessary steps for a do-it-yourself (DIY) project or compose an essay that is certain to deliver exceptional outcomes.
So what really is a ChatGPT prompt?
Getting to Know ChatGPT Prompts
Prompting is the language you speak to your AI; it is a way to tell the AI what you want and how you want it. Typically, a ChatGPT prompt means an instruction or topic of discussion that a user gives to the AI for it to provide a relevant response.
Imagine this: You’re in a situation where you need to create a captivating advertisement for your company’s newest collection of delicate scented candles. What would you do? As if you are hiring a copywriter, the approach remains the same: you express your vision using precise language and clear directions, specifying the needed information such as target audience.
Today’s most powerful AIs function similarly, and naturally, there exists a method to become proficient in generating prompts that yield the most pertinent responses; it is called a Prompt Formula.
What is a prompt formula?
Prompt formulas, also known as prompt engineering, are special techniques that enhance the performance of AI tools, such as ChatGPT. Think of it like giving clear instructions to someone, so they know exactly what you want and can help you faster. Similarly, prompt engineering combines elements of logic, coding, art and special modifiers.
The prompt can include natural language text, images or other types of input data. Although the most common AI tools can process natural language queries, the same prompt will likely generate different results across AI services and tools, which is why it is crucial to better understand your AI assistant, and create a common language between you.
Why You NEED Prompting Formulas
Even though AIs are trained on human-generated data, they are still machines. Hence, generative AI algorithms can produce irrelevant results without the proper context. Prompt engineering helps you get around this issue by providing generative AI models with the proper context and information to produce better results.
Some ways prompt engineering is important include:
Help the AI understand what you want by giving clear instructions for the response. Implementing effective limitations reduces the likelihood of misinterpretations. To achieve this, you must become skilled at controlling your AI.
In other words…
Now how can you become a master at prompting for ChatGPT?
By carefully following step-by-step instructions, you can achieve maximum efficiency when inputting a prompt formula. There are multiple methods available, but with precision and attention to detail, you can ensure exceptional results. We broke the formulas down into two main types as follows:
ChatGPT Prompts 101: Simplified Formulas
This is the formula that requires minimal effort and creativity; being very simple and straightforward, we can break this formula into three main components:
- Action
- Structure
- Direction
In this order, you can type a basic yet efficient prompt that generates a response relevant to your desired output. During this trial we will be using the following as our example:
“Write a blog post about how to start your own business.”
1. Action
The word action here refers to the command you are asking of ChatGPT; it usually should be the first word in your prompt – given that your desired output is a simple and basic one. An example would be: “Write a blog post about how to start your own business.”
You must know exactly what you want your AI to produce as a response; to elucidate, AI works with connotation, so you will need to know what each word means to your AI. That is to say, when you want your AI to write an email perhaps, it could be different from if you command your AI to compose an email. To test that theory, the following was done:
Command: If I tell you to “compose” an email, would it be different from telling you to “write” an email?
AI’s response:In the context of generating text as an AI, the terms “compose” and “write” are often used interchangeably. Both imply the act of generating a written message or document. However, if you want to be more specific or nuanced, you can use “compose” when you want the AI to create a new email from scratch or to emphasize the creative aspect of crafting the email.
On the other hand, “write” can be a more general instruction that includes tasks like drafting, editing, or revising an email. Ultimately, it depends on the level of detail and emphasis you want to convey in your instruction.
- Compose: When you ask me to “compose” an email, it may imply a sense of careful construction, thoughtfulness, and attention to detail. “Composing” often suggests that the email might involve more thought and structure, possibly with attention to the choice of words and overall style.
- Write: On the other hand, when you ask me to “write” an email, it generally refers to the act of putting the content into written form. While it doesn’t exclude thoughtfulness, “writing” might be perceived as a broader term that doesn’t necessarily emphasize the artistic or structured aspects of creating the email.
The previous example only reflects how necessary it actually is for you to know exactly what you are asking out of your AI; your choice of wording can make a world of a difference.
2. Structure
In our trial example, a structure was mentioned: a blog post. The wording here determined what type of written response would be expected as an output. The structure can vary according to your will; it could be an essay, a list, a blog post, or a simple paragraph such as an introductory one.
Once again, wording is a crucial element. You need to have a clear idea of what you want from your AI and give it clear instructions for the helpful result you’re looking for. You can ask it to provide you with the differences between different structures of written responses, so you have a clearer view of what you would want it to provide you with.
3. Direction
The final part of a basic prompt formula is the direction of your command. In other words, it provides your input with some basic context. In the previous example, “how to start your own business” is the direction.
This part could also be referred to as the topic you want to have a response relevant to. You can add more information as to several steps you would like your AI to come up with, or for it to add multiple ways according to different viewpoints, such as intermediate level or something more advanced.
This is also where you can add more context and elevate your input to become a more detailed prompt – which also has its formula!
Using The Right Framework for ChatGPT Prompts
The success of a ChatGPT interaction heavily relies on the clarity and specificity of the prompt. A well-defined prompt provides the model with the necessary context and instructions to generate accurate and relevant responses. Conversely, vague or ambiguous prompts can lead to unpredictable outputs, hindering the user’s experience.
RTF Framework (Role, Task, Format)
The RTF framework provides a structured approach to crafting prompts. By defining the role, task, and format, users guide the model to generate responses tailored to specific needs. For instance:
“Act like a [insert role]. Give me a [insert task] in [insert format] format.”
This framework is particularly effective for tasks that require a specific perspective or format, such as creative writing or problem-solving scenarios.
As an example, you can say: “Act like a content writer. Give me a content schedule for the coming month about digital marketing topics in table format.”
(Use the image of the ChatGPT result from the prompt above.)
CoT Framework (Chain of Thought)
The CoT framework encourages a step-by-step thought process, allowing the model to think through a problem systematically. Utilize this framework to guide the model through a logical sequence of considerations:
“[insert your prompt instructions]. Let’s think through it step-by-step.”
CoT is beneficial for complex inquiries that require a structured and reasoned response, such as providing advice or solving intricate problems.
Pro Tip: If you use a gentler tone with ChatGPT, then you will yield better results from the Generative AI. Using words like “take a breath” or “let’s do this together” will bring better and more creative results from ChatGPT.
RISEN Framework (Role, Instructions, Steps, End Goal, Narrowing)
RISEN is a comprehensive framework suitable for tasks with multiple facets. It breaks down the prompt into Role, Instructions, Steps, End Goal, and Narrowing:
“Role: [insert role.]
Instructions: [insert task instructions.]
Steps: [insert numbered steps to follow.]
End Goal: [insert goal of the output.]
Narrowing: [insert any constraints or limitations].”
This framework is particularly useful for research-oriented prompts or when seeking detailed and specific responses.
ChatGPT Prompts: Expert Unveiling
- Role-playing (AKA: voice/persona)
- Context
- Examples (and/or references)
- Tone/Style
- Setting boundaries (AKA: handholding)
To improve the accuracy and relevance of your response, follow these steps after you have mastered the basic prompt formula.
1. Role-playing (AKA: voice/persona) – PRO Tip!
This one is great for defining a certain voice as background for your AI. This is done by telling your AI to impersonate someone; it could be a famous person or a conceptual identity such as a middle school student. Example: You are Jeff Bezos writing a list of great ways to make fast money.
Another way to do this is by making your AI respond to you as if you are your recipient. Example: I am an employer interviewing a potential employee for the position of a copywriter; write a list of questions I could ask them.
Of course, here, you could be the one being interviewed, but this is a great way to know what they could ask you during the interview.
With some careful prompting, you can get ChatGPT out of its matter-of-fact, default tone and into something much more interesting – such as the style of your favourite author, perhaps.
You could go for the searing simplicity of an Ernest Hemingway or Raymond Carver story, for instance, or the lyrical rhythm of a Shakespearean play, or the density of a Dickens novel. The results don’t come close to the genius of the actual authors themselves, but it’s another way of being more creative with the output you get.
2. Context
- Background
- Environment
- Goal
Providing context allows your AI to have a finer grasp on what you are seeking, and how it can provide the most relevant response. It is one of the most important parts of your formula.
An example would be: I’m a 30-year-old looking to learn English over the next six months. I only have time to do one lesson per day for two hours per lesson, give me a six-month program to follow.
Now context can be broken down into three main parts: background, environmental, and a final goal.
For instance, the background should always be your first contextual input in your prompt; in elaboration, background and environmental context are what steer your AI in the right approach to follow your command.
Environmental context provides your AI with the atmosphere of the desired response; it can be anything you want it to be, and it is crucial to be as clear as possible.
An example would be: I am a business manager sending an email to my marketing executive regarding a business plan that we need to set a meeting for, compose an email and mention my availability for the meeting next Tuesday.
In this example, the part about the meeting for the business plan is considered the environmental context that your AI would need to best understand the desired output it should respond with.
The final part of your context should be the goal you seek by using the response; it could be the result you need itself, the level of complexity you seek, or even the length of the response. In the previous example, the goal was to put on five kilograms of muscle mass, but the options are infinite.
Another example: Create a business plan for a team manager in a company that sells pharmaceuticals to reach a goal of selling 500 pieces in two months.
OR: Write a six-page essay about the history of slavery to submit as a middle school student for a graded school assignment that ensures an expected grade of A.
3. Examples (and/or references)
This is where you provide your AI with a sample of what you want the response to mimic or emulate. You could do that by copying a certain text and asking your AI to write one similar to it.
Example: Write a business plan similar to the one I previously sent.
The criteria to do this is quite simple but requires a bit of focus. You don’t want to have a large sample as input, in order not to confuse your AI; to do so, you can break down your sample into pieces and ask your AI to respond with pieces similar to each one you type in.
An example would be pasting an introductory paragraph, and asking your AI to write one similar to it, then doing the same with the rest of your piece and so on. This can also be done with blog posts, business plans, or any kind of text that can be broken down into pieces.
Another way is to use references such as using a book’s name or a writer’s style.
An example could be: Write the steps for making pesto sauce in the style of a business plan written by Jonah Berger.
4. Tone/Style -Give yourself an edge!
This is where you ask your AI to provide the response in a certain tone that you determine, such as: humorous, playful, or professional. Example: Write an email to be sent to a team of copywriters telling them about the task’s deadline being postponed in a playful tone.
You should make sure you know what each tone entails; for example, there is a difference between a humorous and a playful tone. To elaborate, a humorous tone entices laughter, whereas a playful tone can be used for a carefree output.
Example of input: write me a tweet about the importance of teamwork in a playful tone
Example of input: compose a Facebook ad for a job as a sales rep for a cosmetics company, and make the tone professional.
5. Setting boundaries (AKA: handholding)
Here is where you tell your AI to avoid certain liabilities or ensure they include certain keywords; by doing this, you ensure a higher level of accuracy.
Example: Write an email telling my boss I will be accepting the task they sent and submitting it by the due date; include that a bonus shall be applied.
Now add all of this together!
PROMPT: You are Elon Musk writing a business plan for your new launch of the Tesla vehicle; your business plan should include product and cost sections, a marketing strategy and a sales prediction with the goal of a profit of 35%; it should be written in a professional tone.
Extras:
6. Iterate and refine
Do not hesitate to re-enter the prompt after editing; practice makes perfect. You can ask your AI to provide you with more information.
Example: Set a diet plan for a 35-year-old male who wants to lose 5 kg of fat weight.
AI’s reply could be slightly vague; that is when you iterate and refine.
Example: Set a diet plan for a 35-year-old male who wants to lose 5 kg of fat weight, and include vegetarian options for three meals per week.
You can keep doing so until you get the most relevant response out of your AI.
You can also enable the memory button to make sure your past responses are saved, and you can ask follow-up questions without re-entering the same prompt over and over after updating it.
You should know that prompt engineering is an iterative process. Since there are no rules for how the AI outputs information, you should test different ideas to see what kind of output the AI produces.
You won’t be able to do this in a few tries. You will need to constantly test ChatGPT prompts to see the results and iterate on your discoveries. Ensure accuracy, relevance, and that all other aspects align with your requirements. More optimisation can help you reduce your prompt size to generate better output.
7. Using double quotations
By doing this, you make sure the AI uses the quoted words as focused keywords. If you want to emphasize something, usually you use double quotations in text, or air quotes in real-life conversations. This works with LLMs as well; if you want to make sure your AI focuses on a certain point or word, just use double quotations.
8. Use ChatGPT
Ask your AI to provide you with the perfect prompts and learn as you go by. This technique is parallel to Reverse Prompt Engineering…
So what is Reverse Prompt Engineering?
Reverse prompt engineering is an innovative method that allows you to generate high-quality responses from your AI.
Rather than providing a narrow prompt, this technique starts with the desired response and cleverly works its way back to determine the prompt that would most likely lead to that specific response. This approach is used to understand how a Generative AI generates certain outputs and to uncover the reasoning or patterns behind its responses. -this was ChatGPT’s response!
You can simply paste a text into your chat and ask your AI to provide you with the prompt that would result in that text being the response. By continuously practising, you will gain a deep understanding of how your AI operates, and discover the multitude of effective commands that can be generated from a single prompt.
Imagine asking an artist to portray a renowned artwork, allowing you to have a crystal-clear vision for your future commissions. This way, you will possess the perfect words to vividly describe something that only exists in your imagination.
In other words, you can activate reverse prompt engineering by telling your AI to generate/create a reverse prompt engineering template for a “topic”.
By doing this your AI will be your own prompt generator and will provide you with prompts that result in the most relevant responses.
Conclusion
To put it simply, understanding the art of ChatGPT prompting is the key to unlocking the full potential of artificial intelligence. With the right prompts, you provide your AI with the necessary context and information to deliver better outcomes.
The process begins with a clear and well-structured prompt formula, consisting of three essential components: action, structure, and direction.
Then, for those aiming to become masters of ChatGPT prompting, there is the formula for experts-to-be, which involves detailed context, role-playing, references, tones, and extra boundaries set. It’s important to iterate and refine your prompts, use double quotations to emphasize important keywords, and learn from AI as you go along.
In essence, using ChatGPT prompts is an art that allows you to make the most of AI’s vast abilities. Mastering this skill allows you to use AI for various tasks; this can range from writing to problem-solving.
As AI technology progresses, your knowledge of ChatGPT prompt engineering will be valuable in navigating the AI-driven world. It is important to emphasize that prompt engineering is not limited to ChatGPT. By mastering this skill, you open up a world of possibilities with other AI tools as well.
