Using ChatGPT is quite easy and beneficial, but are you able to get the most out of it? In this article, we will discuss the best practices for ChatGPT. We will talk about how to use ChatGPT more effectively and efficiently, and we will explain how you can achieve better results from it. Our aim with this article is to help you improve your questions or prompts and cover a wide range of topics to maximize the benefits you can get from ChatGPT.
Strategies for using ChatGPT to get better results
ChatGPT leverages the power of language models such as GPT-3.5 and GPT-4. These models has its own unique usage characteristics and best practices. This article shares strategies and tactics for getting better results from ChatGPT. The methods described below can sometimes be used in combination for greater effect.
1. Write Clear Prompts
Writing clear instructions is essential for effective communication and achieving the desired outcomes. Whether you’re interacting with ChatGPTs or providing instructions to humans, clarity is key. We will share some tactics to help you write prompts that are concise, unambiguous, and easily understandable.
ChatGPT can’t read your mind. If outputs are too long, ask for brief replies. If outputs are too simple, ask for expert-level writing. If you dislike the format, demonstrate the format you’d like to see. The less ChatGPT have to guess at what you want, the more likely you’ll get the better results.
Tactic 1: Include Details in Your Prompts to Get More Relevant Answers
When seeking responses or solutions, it’s crucial to provide relevant details and context. By including specific information, you can increase the chances of receiving highly relevant and accurate responses. Without clear details, the ChatGPT might have to guess your intentions, leading to potential misunderstandings. Here are some examples highlighting the difference between vague and detailed queries:
- Bad Prompt: How do I add numbers in Excel?
- Good Prompt: How do I add up a row of dollar amounts in Excel? I want to do this automatically for a whole sheet of rows with all the totals ending up on the right in a column called “Total.”
- Bad Prompt: Summarize the meeting notes?
- Good Prompt: Summarize the meeting notes in a single paragraph. Then write a markdown list of the speakers and each of their key points. Finally, list the next steps or action items suggested by the speakers, if any.
- Bad Prompt: Write code to calculate the prime number in a given array.
- Good Prompt: Write a Python function to efficiently find prime number in a given array of numbers. Comment the code liberally to explain what each piece does and why it’s written that way.
Tactic 2: Ask the Model to Adopt a Persona
You can specify the persona that the ChatGPT should adopt in its replies. By doing so, you can customize the tone and style of the generated content to suit your needs. For instance:
A good prompt example: When I request assistance with writing, your responses will include a document that incorporates at least one humorous joke or lighthearted remark in each paragraph.
Tactic 3: Use Delimiters
Delimiters such as triple quotation marks, XML tags, or section titles can help you demarcate different sections of text, allowing you to treat them differently. They enable you to convey multiple sections within a single query to ChatGPT. For example:
A good prompt example: Summarize the text delimited by triple quotes with a haiku. “””insert your text here”””
Tactic 4: Specify the Steps to Complete a Task
For tasks involving multiple steps, it is helpful to provide explicit instructions that outline each step. This clarity facilitates the ChatGPT’s understanding and improves the accuracy of the generated responses. Here’s an example:
A good prompt example: Follow these step-by-step instructions to respond me: Step 1: I will provide text enclosed within triple quotes. Summarize this text in one sentence with a prefix that says “Summary: “. Step 2: Translate the summary from Step 1 into Spanish, using a prefix that says “Translation: “.
Tactic 5: Provide Examples
Providing examples can be helpful in certain cases, especially when you want ChatGPT to emulate a specific style or response format. This style of prompting allows the ChatGPT to learn from specific examples. For instance:
A good prompt example: Answer my question (Teach me about the ocean) as in the example below. Example question: Teach me about patience. Example answer: Patience is like a river carving the deepest valley, a symphony originating from a single note, or an intricate tapestry starting with a solitary thread.
Tactic 6: Specify the Desired Length of the Response
You can instruct ChatGPT to produce outputs of a specific length, such as word count, sentence count, paragraphs, or bullet points. However, note that specifying the exact number of words may not always be precise. The model can more reliably generate outputs with a specific number of paragraphs or bullet points. For example:
A good prompt example: Summarize the text delimited by triple quotes in 2 paragraphs. “””insert text here”””
2. Utilizing Reference Text for Improved Answers
ChatGPT possess the capability to generate fake responses. Especially when confronted with obscure subjects or requests for citations and URLs. However, we can employ reference text strategically to minimize such fabrications and provide more reliable answers.
Tactic 1: Instruct ChatGPT to answer using a reference text
We can instruct ChatGPT to use the provided information to prepare its answer if we can provide it with trusted information that is relevant to our question.
A good prompt example
Use the provided articles delimited by triple quotes to answer questions. If the answer cannot be found in the articles, write "I could not find an answer." """insert articles here, each delimited by triple quotes""" Question: <insert question here>
Tactic 2: Instruct ChatGPT to answer with citations from a reference text
We can request ChatGPT to add citations to its answers by referencing passages from provided documents if the input has been supplemented with relevant knowledge. It’s worth noting that the citations in the output can be programmatically verified by string matching within the provided documents.
A good prompt example
You will be provided with a document delimited by triple quotes and a question. Your task is to answer the question using only the provided document and to cite the passage(s) of the document used to answer the question. If the document does not contain the information needed to answer this question then simply write: "Insufficient information." If an answer to the question is provided, it must be annotated with a citation.
Use the following format for to cite relevant passages ({"citation": …}).
"""<insert document here>"""
Question: <insert question here>
3. Split complex tasks into simpler subtasks
In software engineering, it’s common practice to break down a complicated system into smaller, independent parts called modules. Similarly, when using GPTs, it’s beneficial to divide complex tasks into simpler ones. Complex tasks often have more errors, while simpler tasks are more manageable.
Therefore, when submitting questions to ChatGPT, you should follow the same strategy.
Tactic 1: Utilize intent classification
When dealing with tasks that involve handling different cases, it is beneficial to first classify the type of query. This classification can help identify the specific instructions required. One way to accomplish this is by establishing fixed categories and including relevant instructions for each category. By doing so, the task can be broken down into stages using a recursive approach.
The advantage of utilizing intent classification is that it allows each query to include only the instructions needed for the next stage. This targeted approach reduces the likelihood of errors compared to attempting the entire task in a single query.
For example, let’s consider a customer service application. Queries in this context can be effectively classified based on different categories. This classification approach enables the system to provide the appropriate instructions for handling tasks within each category.
A good prompt example:
You will be provided with customer service queries. Classify each query into a primary category and a secondary category. Provide your output in json format with the keys: primary and secondary. Primary categories: Billing, Technical Support, Account Management, or General Inquiry. Billing secondary categories: Unsubscribe or upgrade Add a payment method Explanation for charge Dispute a charge Technical Support secondary categories: Troubleshooting Device compatibility Software updates Account Management secondary categories: Password reset Update personal information Close account Account security General Inquiry secondary categories: Product information Pricing Feedback Speak to a human <User Question in the customer service: I need to get my internet working again.>
By utilizing intent classification, the system can identify the category of the customer query, such as “troubleshooting”. This classification enables the system to provide tailored instructions specifically designed to address troubleshooting-related issues. These instructions can be passed to the ChatGPT. This will guide it to generate responses that accurately assist the customer in resolving their problem.
Following good prompt example
You will be provided with customer service inquiries that require troubleshooting in a technical support context. Help the user by:
Ask them to check that all cables to/from the router are connected. Note that it is common for cables to come loose over time.
If all cables are connected and the issue persists, ask them which router model they are using - Now you will advise them how to restart their device:
If the model number is MTD-327J, advise them to push the red button and hold it for 5 seconds, then wait 5 minutes before testing the connection.
If the model number is MTD-327S, advise them to unplug and replug it, then wait 5 minutes before testing the connection.
If the customer's issue persists after restarting the device and waiting 5 minutes, connect them to IT support by outputting {"IT support requested"}.
If the user starts asking questions that are unrelated to this topic then confirm if they would like to end the current chat about troubleshooting and classify their request according to the following scheme:
<insert primary/secondary classification scheme from above here>
<User Question in the customer service: I need to get my internet working again.>
Tactic 2: For very long conversations, summarize or filter previous dialogue
While ChatGPT can remember previous parts of a conversation and continue from there, it has a character limit in a conversation window. If the conversation becomes too long, ChatGPT may not be able to recall the initial details you provided. To assist ChatGPT in such cases, if the conversation extends significantly, it would be helpful if you could provide a brief summary of the key points from your previous messages. This will ensure that ChatGPT has the necessary context to continue the conversation smoothly.
Tactic 3: Summarize long documents part by part
While ChatGPT is capable of summarizing text, it does have a limitation on the character limit. Therefore, when requesting a summary of a lengthy text, ChatGPT may not be able to summarize the entire content in one go. To overcome this limitation, it is advisable to break down the lengthy text into smaller parts and request summaries for each section separately. This way, ChatGPT can provide more accurate and concise summaries for each segment of the text.
4. Give ChatGPT some time to think
When you face with a question like multiplying 17 by 28, you might not have an immediate answer. However, you can work it out with some time. Similarly, ChatGPT tends to make more errors when trying to provide instant answers rather than taking the time to reason through the problem. By asking for a chain of reasoning before expecting an answer, ChatGPT can engage in a more thoughtful process and improve its ability to arrive at correct answers more reliably.
Tactic 1: Instruct ChatGPT to work out its own solution
Sometimes, instructing the model to reason from basic principles before reaching a conclusion can lead to improved outcomes. For instance, when evaluating a student’s math solution, instead of directly asking the model if it’s correct, we can guide it to consider the underlying principles first. So we should apply the same principle here.
A good query example
First work out your own solution to the problem. Then compare your solution to the student's solution and evaluate if the student's solution is correct or not. Don't decide if the student's solution is correct until you have done the problem yourself. Problem Statement: I'm building a solar power installation and I need help working out the financials. Land costs $100 / square foot I can buy solar panels for $250 / square foot I negotiated a contract for maintenance that will cost me a flat $100k per year, and an additional $10 / square foot What is the total cost for the first year of operations as a function of the number of square feet. Student's Solution: Let x be the size of the installation in square feet. Land cost: 100x Solar panel cost: 250x Maintenance cost: 100,000 + 100x Total cost: 100x + 250x + 100,000 + 100x = 450x + 100,000
Tactic 2: Ask ChatGPT if it missed anything
ChatGPT is a helpful tool that can assist us in understanding long documents. However, sometimes it might miss important points while providing us with information. That’s why it can be useful to ask ChatGPT if it missed any important points after it gives us an answer. This way, if ChatGPT skipped or overlooked anything, it can let us know.
5. Use External Tools
ChatGPT has many useful capabilities that can benefit us in various ways. However, there is no reason to limit ourselves to just ChatGPT. We can enhance our productivity by leveraging external tools to compare and augment the answers provided by ChatGPT. For example, we can use the Grammarly application to check the grammar accuracy of the responses given by ChatGPT. This allows us to automatically scan and ensure correctness. Similarly, we can utilize various external third-party tools like ChatGPT plugins to enhance its existing capabilities and derive additional benefits.
How to write faster on ChatGPT?
I understand that sometimes it can be tedious and time-consuming to write detailed prompts or questions while using ChatGPT. However, there’s a great method you can use called voice typing in ChatGPT. It allows you to speak your prompts instead of typing them out, which can save you time and effort.
You can indeed use the voice typing feature in ChatGPT to prepare your prompts by speaking to the model instead of typing. This allows you to provide detailed information about the topic you want to discuss, including all the necessary details and specifics. By doing so, you can accurately express the problem or issue you want to address with ChatGPT without having to spend too much time on it. So go ahead and utilize the voice typing feature to save time while effectively communicating your thoughts to ChatGPT.
To accomplish this, all you need to do is use the “ChatGPT Voice Typing” Chrome extension. It enables you to utilize the voice typing feature in ChatGPT. Simply install the extension in your Chrome browser, and you’ll be able to speak your prompts and questions directly to ChatGPT
Leave a Reply