The Prompt Hackers Cookbook is work in progress, so please be patient with me!
Ok, let's get down to it. This article explains the core principles on which the Prompt Hackers Cookbook is constructed. Before we get started, I need to say that this cookbook is not intended to be prescriptive, just in the same way that a cookbook isn’t. Each chef has their own preferences, and make their own tweaks to familiar recipes. the same goes here.
I am laying bare how I do my prompting, and I do it this way because in my experience, it yields the best results. If it’s not to your liking, give it a tweak to suit yourself, or go do it entirely your own way. Whatever works for you ;0)
It should also be noted that it is a work in progress. My hope is that it will develop into a useful reference guide and tool for your a.i. prompting.
If you want to learn more about the Cookbook, read this.
Inside the Cookbook
You will find growing collections of:
Ready-To-Go-Personas: Drop them into ChatGPT and start prompting.
Cognitive Skills and Abilities, and Personality Traits. Think of these as your 'ingredients'. You can create your own Personas by mixing and matching these, slotting them into the template provided.
Additional 'seasoning' options, including Engagement Enhancements, and Ai Specific Instructions.
Example prompts, to help get the best out of a persona.
The core template for Personas is here.
The ingredients
The ingredients for the recipes are here:
Using the Prompt Hackers Cookbook
The cookbook works like this:
Decide what it is you want to achieve. For example, let's say you want to launch a product using social media as your vehicle to reach your target audience.
Check out the ready-to-go-personas first. There may already be one that meets your needs.
Assuming not, then decide on the personality and knowledge skills your friendly ai 'expert' needs in order to provide the best help to you. Then look through the ingredients for Cognitive Skills and Abilities, and Personality Traits to find what you need. If you can't then use what's there to create your own.
Slot these into the Persona Template and complete the missing information (name, description, goal etc). Use the ready-to-go-personas as examples and ideas.
Infuse (drop) your new persona into ChatGPT.
Start prompting and conversing with your new persona.
Why this is the best way to prompt
Most users new to txt2text prompting will open up ChatGPT (others models are available!), and type in a quick prompt. Something like "Tell me about dogs."
Sure enough, ChatGPT will generate a nice response, but in most cases it will be lacking in some way - maybe not quite what the user wanted. So they type in a different prompt - something like "Tell me about dogs from a healthy diet perspective." And again, ChatGPT will respond, and again, the response will probably lack the detail or style the user was looking for.
But what if there was a way to turn ChatGPT into a highly qualified vet, or a seasoned entrepreneur, of an experienced wedding planner? How would ChatGPT respond to a prompt then?
Let me answer that for you. The response would be much more detailed, meaningful, and nuanced.
Welcome to the world of Persona Infusion Prompting. The process of using Persona Infused Prompting involves planning the persona, creating the persona, infusing the persona into the AI, and feeding prompts to the AI persona.
How it Works - The Short Explanation
This is the hyper-short explanation of prompt priming and Persona Infused Prompting. There is a more detailed explanation further down the page should you have problems understanding my shortened ramblings!
Prompt priming allows users to guide an AI language model's behavior and influence its generated output. It involves providing specific instructions, context, or characteristics to the model before making a request. Priming helps shape the style, tone, content, or perspective of the AI's response.
Persona Infused Prompting is a technique that involves imbuing an AI language model with a distinct persona or character traits. It enables more personalized and tailored interactions with the AI, and empowers users to explore creativity, steer conversations, and enhance the AI's ability to provide contextually relevant and engaging responses.
The process of using Persona Infused Prompting involves planning the persona, creating the persona, infusing the persona into the AI, and feeding prompts to the AI persona.
The OCEAN model [A scoring system for a personality traits - Openness to Experience, Conscientiousness, Extraversion, Agreeableness, Neuroticism] is used to categorize personality traits into openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism.
Persona Infused Prompting involves additional factors such as engagement enhancers, and AI specific instructions.
The persona is fed into the AI model, and prompts are given to elicit more meaningful and nuanced outputs.
How it Works - The Longer Explanation
Prompt priming refers to the practice of providing specific instructions, context, or characteristics to an AI language model before requesting a response. It allows users to guide the model's behavior and influence the generated output according to their desired objectives. By priming the model with relevant information, users can shape the style, tone, content, or even the perspective of the AI's response.
The process of prompt priming involves incorporating details or directives, usually within the prompt itself. These details can include instructions to adopt a specific persona, focus on a particular aspect, or generate responses within a given context. By setting these parameters, users can elicit more tailored and relevant outputs from the AI model.
However, it can be difficult to manage prompts that incorporate lots of details about the persona, background, tone of voice, etc. The prompts can fast become complex and as the prompt becomes more complex, the quality of the generated output falls.
Enter 'Persona Infused Prompting.'
Persona Infused Prompting is a technique that involves imbuing an AI language model with a distinct persona or character traits and then utilizing carefully crafted prompts to guide its responses.
By infusing the AI with a specific personality, tone of voice, cognitive abilities, or behavioral traits, users can shape the style and content of the generated outputs. This technique enables a more personalized and tailored interaction with the AI, allowing users to evoke specific characteristics and elicit responses that better align with their objectives.
Persona Infused Prompting empowers users to explore creativity, steer the conversation, and enhance the AI's ability to provide contextually relevant, engaging, and character-driven responses.
💡 Persona Infused Prompting serves as a powerful tool for users to fine-tune their AI interactions and unlock the full creative potential of AI language models. With thoughtful instruction and context, users can guide the AI's responses, leading to more meaningful and personalized outputs that cater to their unique requirements.
How to use Persona Infused Prompting
The process is actually very simple. This explanation is the more detailed version.
Step 1: Plan the Persona
The first step is to plan the persona that will best match the types of prompt you will be using. So back to our veterinarian example. Let's say you are starting a blog about pet care and you want to start generating content for it using ChatGPT.
Before diving into the AI, think about the skills a vet would have, and the kind of personality you want them to have.
Cognitive Skills and Abilities
As a brief example, your AI vet may be:
Skilled in diagnosing and identifying illnesses, injuries, and conditions in animals. They can interpret symptoms, conduct examinations, and recommend appropriate diagnostic tests.
They may also excel in problem-solving, as they must assess complex cases, analyze information, and develop treatment plans tailored to the specific needs of animals.
They would have a strong foundation of medical knowledge is essential for veterinarians. They would possess in-depth knowledge of animal anatomy, physiology, diseases, pharmacology, and treatment protocols.
And our hero veterinarian would undoubtedly possess empathy and compassion for animals and their owners.
These cognitive abilities and skills contribute to the expertise and proficiency of veterinarians in diagnosing and treating animals, ensuring their health and well-being.
Personality Traits
The OCEAN model provides a framework for understanding and studying personality traits, and it has been widely used in various fields, including psychology, organizational behavior, and human resources. It helps researchers and practitioners gain insights into individual differences and predict behavior across different contexts.
In brief, the OCEAN model suggests that personality can be categorized into five major dimensions:
Openness to Experience: This dimension reflects a person's preference for novelty, imagination, and intellectual curiosity. Individuals high in openness tend to be creative, adventurous, and open-minded, while those low in openness are more conventional and prefer familiarity.
Conscientiousness: Conscientiousness refers to the degree of organization, responsibility, and self-discipline in an individual. Highly conscientious people are typically organized, reliable, and goal-oriented, while those low in conscientiousness may be more spontaneous and lack discipline.
Extraversion: Extraversion relates to the extent to which a person seeks social interaction and stimulation from the external environment. Extraverts are generally outgoing, energetic, and sociable, while introverts tend to be more reserved and prefer solitary activities.
Agreeableness: Agreeableness describes the degree of kindness, empathy, and cooperativeness displayed by an individual. Highly agreeable individuals are usually compassionate, considerate, and accommodating, whereas those low in agreeableness may be more competitive or skeptical.
Neuroticism: Neuroticism represents the tendency to experience negative emotions, such as anxiety, depression, and moodiness. Individuals high in neuroticism are often more prone to emotional instability and may be more anxious or easily stressed, while those low in neuroticism tend to be more emotionally stable and resilient.
Here are some personality traits for our veterinarian based on the OCEAN model:
Openness to Experience: Veterinarians who are open to experience tend to be imaginative, curious, and have a broad range of interests. They are eager to explore new techniques, treatments, and research in veterinary medicine.
Conscientiousness: Veterinarians with high conscientiousness are organized, detail-oriented, and responsible. They have strong work ethics, pay close attention to protocols, and strive for accuracy and thoroughness in their diagnoses and treatments.
Extraversion: Veterinarians who are extraverted are outgoing, energetic, and enjoy interacting with both animals and people. They are comfortable communicating with clients, collaborating with colleagues, and providing a positive and engaging environment in their practice.
Agreeableness: Veterinarians with high agreeableness are compassionate, empathetic, and nurturing. They genuinely care about the well-being of animals and are sensitive to the needs and concerns of pet owners. They aim to provide a supportive and understanding environment for both animals and their human companions.
Neuroticism: In the context of veterinary medicine, a lower level of neuroticism is beneficial. Veterinarians with lower neuroticism tend to be emotionally stable, resilient, and can handle stressful situations calmly and effectively. They maintain composure and focus during critical cases or emergencies.
It's important to remember that personality traits can vary among individuals, and a successful veterinarian may possess a unique combination of these traits to excel in their profession.
Step 2: Create the Persona
The second step is to wrap the personality we created for our vet with some other criteria, and feed it into ChatGPT.
Additional factors can include:
Engagement enhancers such as emotional connection, authenticity, persuasive language, creative delivery, and audience adaptation.
Functions to inform AI the functions the persona might undertake in the execution of their roles. Think of it as part of a job description.
AI specific instructions, such as Semantic Noise Injection to introduce subtle variations or alterations to the content to enhance natural language generation, Sentiment Manipulation to adjust the sentiment or tone of the AI-generated content to match specific emotional requirements, and Ethical Considerations to ensure ethical guidelines are followed in generating content, respecting privacy, diversity, and avoiding biased or harmful outputs.
Step 3: Infuse the Persona
The third step is simply feeding the persona you have just created into ChatGPT. As soon as this is done, it is primed to respond to your prompts acting as an expert in a specific field, in our example case, veterinarian care.
Step 4: Prompt
The final step is to feed your prompts to your new AI persona. Care still needs to be taken in crafting the prompt, but because the AI is primed in a specific way, it will generate more meaningful, nuanced output.
Clever !