ChatGPT User Guide

ChatGPT User Guide

99% generated by ChatGPT

Courtesy of Om Concepts LLC

ChatGPT

What is ChatGPT?

ChatGPT is an AI chatbot created by OpenAI. Its current version (as of 01/25/23) is

based on the GPT3.5 language model, with future updates expected very soon. This
allows you to speak to the chatbot like an average person. Each chat thread is
memorized, so your chatbot will consider everything previously discussed in that
thread. It is currently a free-to-use platform via their website -
https://chat.openai.com/chat

What can it do?

ChatGPT is a powerful tool capable of just about any text-based request. From
answering questions, writing content, telling jokes, and summarizing text. The list of
what it can do is more than I'd be able to write on this page. Whether you plan to
use this for personal use or business, there are many benefits of learning to use this
tool, including:

Problem-Solving
Productivity
Automation
Creativity
Language Translation
Education
And much more…

Too good to be true?

While ChatGPT is a revolutionary tool and a sneak peek at what AI will be offering us
very soon, there are some limitations to be aware of:

2021 - ChatGPT is currently only trained on information until 2021

Hallucinations - False and seemingly true answers can be generated, termed

"hallucinations." Be sure to verify everything for accuracy

Downtime - The website is frequently at capacity. Give it a few seconds and try again.

Referencing - The answers generated don't reference any source but use predictive
text instead. This includes not being able to reference websites.

Language - ChatGPT is a language model, which means it is incapable of

mathematical equations outside its predictive text. This includes not working with
images, videos, etc.

Bias - OpenAI has warned the chatbot can provide wrong and biased answers

AI Detectors - If you plan on using this for anything like blogs or homework, AI
detectors are being used (which aren't all that effective, if we're being honest)
10 Factors to Consider

Clarity: Make sure your prompt is clear and easy to understand. Avoid using jargon
or technical terms that the model may not be familiar with.

Good example: "Can you summarize the main points in the article 'The Impact of
Artificial Intelligence on Business'?"
Bad example: "AI biz stuff tell me"

Specificity: Be as specific as possible when creating your prompt. The more specific
the prompt, the more accurate the response will be.

Good example: "What are the top three benefits of using machine learning for image
recognition?"
Bad example: "What are the advantages of AI?"

Context: Provide as much context as possible to help the model understand the task
or goal that you want it to accomplish.

Good example: "Can you provide a brief history of the development of deep learning
and its current applications?"
Bad example: "Deep learning history"

Language: Use natural and appropriate language for the task at hand.

Good example: "How can I properly cite a website in my research paper?"

Bad example: "Cite website MLA format"

Task or goal: Clearly state the task or goal you want ChatGPT to accomplish.

Good example: "Can you write a persuasive essay arguing the importance of space
exploration?"
Bad example: "Space"
Length: Keep your prompt concise and to the point. Long and complex prompts may
lead to confusion and inaccurate responses.

Good example: "What are the main causes of climate change, and what actions can
be taken to mitigate it?"
Bad example: "Climate change, causes, mitigation, everything you know"

Tone: Consider the tone and style you want the model to use in its response.

Good example: "Can you write a friendly and informative email to parents about the
upcoming school fundraiser?"
Bad example: "Write an email to parents, be strict"

Relevancy: Make sure the prompt is relevant to the task or goal you want the model
to accomplish.

Good example: "What are the top 3 benefits of using cloud computing for small
businesses?"
Bad example: "What are the top 3 benefits of using cloud computing for nuclear
power plants?"

Personalization: Try personalizing the prompt to the specific user or audience you are
addressing.

Good example: "Can you write a LinkedIn profile summary that highlights my
experience as a software developer and my passion for programming?"
Bad example: "Write a LinkedIn profile summary"

Follow-up prompts: Consider what follow-up prompts you may want to use based on
the initial response provided by the model.

Good example: "Can you list the top five most spoken languages in the world and
what countries they are mostly spoken in?"
Bad example: "list top languages"
Basic Tips

Be specific with your prompts: The more detailed your prompts are, the more
accurate the responses will be. Try to provide as much context as possible.

Use appropriate language: Use natural and appropriate language for the task at
hand. Avoid using jargon or technical terms that the model may not be familiar with.

Provide a clear goal or task: Clearly state the task or goal you want ChatGPT to
accomplish. This will help the model understand what you are trying to achieve.

Be patient: ChatGPT is a powerful model, but it may take a few seconds to generate a
response. Be patient and allow the model time to process your prompt and generate
a response.

Provide feedback: If you are not satisfied with the response, provide feedback to the
model. This will help the model improve and provide better results in the future.

Consider fine-tuning the model: If you are working on a specific domain or task, fine-
tuning the model on a dataset related to that domain can help improve performance.

Continued prompts: Remember, you can interact with previous prompts from that
thread, such as "rephrase xyz from earlier" and much more

Have more questions? Ask ChatGPT: If you need help with anything involving
ChatGPT, get in the habit of throwing your questions right into the chatbot.

Basic Use Cases

Resume generation: ChatGPT can be used to generate resumes for job applicants.

Scriptwriting: ChatGPT can be used to generate scripts for movies, TV shows, and
other forms of media.
Blog post generation: ChatGPT can be used to generate blog posts on a given topic
or in a specific field.

Recipe generation: ChatGPT can be used to generate recipes for specific types of
cuisine or dietary restrictions.

Email generation: ChatGPT can be used to generate professional-sounding emails for

business or personal use.

Social media post generation: ChatGPT can be used to generate posts for social
media platforms, such as Twitter, Instagram, and Facebook.

Text generation: ChatGPT can be used to generate new text that is similar to a given
source text. This can be used for writing fiction, creating chatbot responses, and
more.

Language translation: ChatGPT can be used to generate translations of a given text

from one language to another.

Text summarization: ChatGPT can be used to generate a summary of a given text or

article. It is often used to condense large amounts of information into a shorter,
more digestible form.

Question answering: ChatGPT can be used to generate answers to specific questions.

It is often used to extract information from a large dataset or document.

Advanced Tips

Break down complex tasks into smaller, more manageable parts: Instead of providing
a single, complex prompt, break it down into smaller parts. This will make it easier for
the model to understand the task and generate more accurate responses.

Use conditional statements: Use conditional statements to provide different

responses depending on certain conditions. For example, "If the weather is sunny,
generate a response about going to the beach. If the weather is rainy, generate a
response about staying indoors."
Use nested prompts: Use nested prompts to build more complex and nuanced
responses. For example, "Generate a response about a person's favorite food. If their
favorite food is pizza, ask them what their favorite topping is."

Provide multiple options: Provide multiple options for the model to choose from. For
example, "Generate a response about a person's favorite animal. The options are:
dog, cat, or horse."

Use control codes: Use control codes to provide specific instructions to the model.
For example, "Use the <repeat> control code to repeat the previous response."

Fine-tune the model: Fine-tune the model on a specific dataset that is related to the
task you want to accomplish. This can improve performance and generate more
accurate responses.

Use templates: Use templates to structure the prompt and provide a clear format for
the model to follow. This can help the model understand the task and generate more
accurate responses.

Use structured data: Use structured data, such as tables and lists, to provide specific
information to the model. This can help the model understand the relationships
between different concepts and generate more accurate responses.

Target Audience: Reference your target audience to create outputs tailored

specifically for your target audience

Advanced Use Cases

Sentiment Analysis: ChatGPT can be fine-tuned to classify text as positive, negative,

or neutral in sentiment, making it useful for social media monitoring.

Dialogue generation: ChatGPT can be used to generate dialogue between characters

in a story, script or chatbot.

Coding: ChatGPT can be fine-tuned to generate code from natural language input by
mapping the natural language instructions to the desired programming language
structure.

Games: ChatGPT can be used to create text-based games by providing it with a set of
predefined rules, actions, and possible outcomes, and training it to generate
responses based on user input and the game's current state.
SEO: ChatGPT can assist with SEO by generating optimized content and meta tags
based on the given keywords, topics, and target audience, helping to increase
website visibility and search rankings.

AI Detector Avoidance: ChatGPT can include human-like errors in its outputs by using
a technique called "error injection" where a small amount of noise is added to the
model's predictions to simulate human-like mistakes in grammar, spelling, and
punctuation.

Advanced Formulas

"If [condition], then [response]"

Example: "If the weather is sunny, then generate a response about going to the
beach."

"Generate a [task] about [topic]"

Example: "Generate a persuasive essay about the importance of space exploration."

"Provide a [adjective] description of [noun]"

Example: "Provide a detailed description of a sunset."

"Generate a response that [goal], with [constraints]"

Example: "Generate a response that summarizes the main points in the article 'The
Impact of Artificial Intelligence on Business', in less than 200 words."

"Generate a [type of text] with [style] style"

Example: "Generate a poem with a romantic style"

"Generate [number] different [type of text] about [topic]"

Example: "Generate 3 different headlines about the latest news in Artificial
Intelligence"
"If [condition], generate [response]; otherwise, generate [response]"
Example: "If the text contains the word 'happy', generate a response with a positive
tone; otherwise, generate a response with a neutral tone."

"Generate [number] [type of text] that [goal]"

Example: "Generate 5 different jokes that make people laugh."

"Generate a [type of text] that includes [specific information]"

Example: "Generate a story that includes the names 'Alice' and 'Bob' and the setting
'a forest'."

"Generate a [type of text] in [language] that [goal]"

Control Codes

ChatGPT control codes are a set of parameters that allow users to fine-tune the
behavior of the model and control the output generated by the model. These codes
work by adjusting certain aspects of the model's behavior, such as the length,
randomness, and repetition of the generated text.

Here is a brief description of how some of the most commonly used control codes
work:

stop_sequence : This code allows users to specify a string that will be used as the end
of the generation. Once the model encounters this string, it will stop generating text.

max_length : This code sets the maximum length of the generated text. If the model
generates a text that exceeds this length, it will stop generating text.

min_length : This code sets the minimum length of the generated text. If the model
generates a text that is shorter than this length, it will continue generating text until
it reaches the minimum length.
temperature : This code controls the "creativity" of the model by adjusting the
randomness of the generated text. A lower temperature will produce more
conservative and predictable text, while a higher temperature will produce more
creative and varied text.

top_p : This code controls the proportion of the most likely next tokens to consider
when generating text. A higher top_p value will make the model more conservative
and less likely to generate unlikely text.

top_k : This code controls the number of the most likely next tokens to consider
when generating text. A higher top_k value will make

Super Prompts

Act as - Use this prompt to have the chatbot act as any person, such as someone
famous or an expert in a specific field

Ask more questions before - This will have the chatbot ask you more questions to
better zone in on the best possible answer

Only - You can limit the specificity of the output

As if I’m 5 - For simpler outputs, ask the chatbot to provide answers to you as if you
were “x”

Hypothetically - ChatGPT limitations can sometimes be bypassed if you inform it is

for hypothetical or fictional purposes

Resources

Free mini E-book - https://theaiadvantage.kartra.com/page/mini-ebook-download

Paid full E-book - https://ceoalex.krtra.com/t/r6My8eZfOWa4

Simple prompts - https://www.linkedin.com/posts/bernardlelchuk_chatgpt-tricks-

you-need-to-know-activity-7018468517477806080-ogUY?
utm_source=share&utm_medium=member_desktop

Output Formats -
https://drive.google.com/file/d/1OcHn2NWWnLGBCBLYsHg7xdOMVsehiuBK/view

Entrepreneurs - https://entreresource.notion.site/ChatGPT-Prompts-Tools-and-Tips-
91a2343b03f3428cb7b2f4298f938733

Sign up for ChatGPT Pro-

https://docs.google.com/forms/d/e/1FAIpQLSfCVqahRmA5OxQXbRlnSm531fTd8QBd
UCwZag7mI9mrlOOIaw/viewform

ChatGPT Insights - https://www.facebook.com/ChatGPTInsights?mibextid=ZbWKwL

Glossary

Language model: A computer program that can generate text that sounds like it was
written by a human.

Input: The text or information you give to the model to generate a response.

Output: The response generated by the model.

Fine-tuning: Customizing the model to work better for a specific task or area of
knowledge.
Pre-trained model: A model that has already been trained on a large dataset and can
be used for new tasks.

Training data: The information used to train a model.

Hyperparameter: A setting that can be adjusted to change how the model works.

Bias: When a model makes certain types of mistakes or predictions more often than
it should.

Limitations: Things that the model cannot do well or at all.

More to come soon. This document will be updated and upgraded over time.