Using ChatGPT is an art. If everyone knew how to get effective answers from ChatGPT, the world wouldn’t be hungry for prompt engineers, right?
Asking ChatGPT the right set of questions, putting guardrails around AI, and training it to answer in your style and tone isn’t everyone’s cup of tea.
And that’s why we have come up with the 15 most effective ChatGPT tips you must know about if you use ChatGPT daily.
What are the Best ChatGPT Tips and Tricks?
This article has three sections:
- ChatGPT tips from Team-GPT: Our expert team has put together some tested tips to help you get better answers from AI.
- ChatGPT tips from Reddit: Comprehensively researched and tested tips from Reddit users.
- ChatGPT tips in general: A few best practices when using ChatGPT.
What are the Best Team-GPT Tips?
Let’s begin with ChatGPT tips from real experts in the field; that’s our team. 😉
1. Tailoring Prompts and Audience Specification
When instructing ChatGPT, customize your prompts for specific needs, knowledge levels, and audiences. This helps you get appropriate and relevant answers from the AI.
Suppose you’re addressing a professional audience; you’d use technical language and industry-specific terms. But, on the contrary, you will use simpler language when your audience is beginners.
When you set a persona for AI, you are directing it to adopt a certain role or perspective. This helps the AI adapt to the tone and nature of the expert you want it to be.
Prompt: Persona: Act like a seasoned financial analyst.
Goal: Explain the concept of blockchain technology to a group of high school students.
Context: Please provide a simplified overview that includes key principles but avoids overly technical jargon.
Chat link: https://chat.openai.com/share/7e94ede0-1494-4625-b133-1c0dbfadac42
2. Encourage ChatGPT to Ask Clarifying Questions
When you ask ChatGPT to ask clarifying questions to make things clearer, you are expected to get much more precise answers after having a conversation with artificial intelligence.
Also, remember to use clear, direct language and avoid negative constructs like ‘don’t.’ The AI might get confused when you use don’t in your prompts – use do instead.
Prompt: Goal: Generate a business plan for the launch of my company’s new product.
Context: The product is a new type of shoe that is flexible and affordable. Ask me all the clarifying questions you need to understand the product and target market thoroughly. I will answer them for you.
Chat link: https://chat.openai.com/share/bb9af44d-b206-4d39-b805-e8f2c9eec082
3. Enhance Prompt Effectiveness with Chain of Thought Prompting
Chain of Thought (CoT) prompting encourages large language models to explain the reasoning behind their answers. To do so, you must give an example where you explain the reason behind the answer and then follow the reasoning with the next questions.
Let me explain to you with an example:
Prompt: Question: Dave has 5 oranges. He buys 2 more packs of oranges. Each pack has 6 oranges. How many oranges does Dave have now?
Answer: Dave started with 5 oranges. 2 packs of 6 oranges each is 12 oranges. 5 + 12 = 17. The answer is 17.
Question: A fruit vendor has 40 bananas. Suppose he sold 22 in the first half and bought 40 more bananas later. How many bananas is he left with?
Chat link: https://chat.openai.com/share/8231c7d8-c041-4ec5-90c4-f12f82ffb78e
With CoT prompting, we basically show ChatGPT how to answer a specific question.
4. Specify the Tone and Style You want the AI to Answer In
Specify the desired tone using adjectives (e.g., ‘serious,’ ‘funny’) and maintain the style of communication. For text revisions, instruct the AI to keep the original style intact while improving grammar and vocabulary.
Prompt: [insert text]
Goal: Revise the following text to enhance its grammar and vocabulary, ensuring that the original style is preserved.
Context: The text should be modified to sound more professional yet maintain its light-hearted and engaging tone. Additionally, ensure that the revised text flows smoothly and clearly conveys its message.
Chat link: https://chat.openai.com/share/94a31da2-4bba-4203-bf9d-741670e0fa8f
5. Structure Your Prompts Right
At Team-GPT, we have an effective, prompt structuring method that gives you better, precise answers every time you use ChatGPT.
We believe that every prompt must have at least 2 parts:
- Context: A section where you provide ChatGPT all the essential information needed for it to answer the questions you ask.
- Goal: The question you want to answer.
So, if you want ChatGPT to give you title ideas for a specific blog you’re about to write. Your goal is to Generate 10 title ideas for the blog post I am planning to write on [insert keyword]
Your context will be all about what you’re planning to write for the topic, a brief description of the headers you plan to cover, the tone you will write it, etc.
Apart from these 2 parts, there’s one other section we like to add to the prompts, and that’s Persona.
Giving ChatGPT persona, such as – act as a finance expert, act as a seasoned copywriter, etc., will get you better and more specific answers.
Prompt: Persona: Act as a native English copywriter in the Finance industry.
Context: I am planning to write an article on How Businesses Can Save Costs by Reducing Operational Costs.
Goal: Start by defining operational costs for me.
Chat link: https://chat.openai.com/share/fd122eb0-a769-4fa9-8b36-a666d853e534
What are the Best ChatGPT Tips from Reddit?
Reddit is home to curious users who love experimenting and digging for tips, tricks, and hacks to use ChatGPT to the fullest. Below are five of the best ChatGPT tips from Reddit.
6. Reduce Ambiguity
Ambiguity is the quality of being open to more than one interpretation. No matter how good it sounds, when it comes to AI, ambiguity is a big NO. It confuses the chatbot, makes it hallucinate, and ends up giving generic or irrelevant responses.
At times, ChatGPT can also reflect ambiguity. However, Reddit users have a trick to prevent the AI from going haywire.
Here’s how:
To reduce ambiguity and improve the response accuracy, ensure your prompts are as specific as possible.
For example, if you wish to generate a number-specific response, like a list of three, make sure your prompt conveys it explicitly.
Likewise, accompany every prompt with enough context and background. This prevents the AI from hallucinating or going off-track.
When you define specificities in your prompts, you’re asking ChatGPT to follow a predefined path and explore answers in a certain direction. As a result, getting more relevant and targeted responses.
Prompt: Goal: Provide a detailed comparison of the educational systems in Finland, Japan, and Canada.
Context: Specifically, focus on three aspects for each country: the structure of the school system (including primary, secondary, and higher education), the teaching methodologies employed, and the assessment and evaluation strategies used.
Also, include how these elements contribute to the overall success of students in international assessments like PISA. Please ensure the information is up-to-date as of 2024.
Chat link: https://chat.openai.com/share/c945736d-3eea-447e-b26f-6f89f0c732c5
7. Breakdown Complex Questions to Smaller Ones
ChatGPT is a powerful tool. However, it struggles to answer complex questions. Especially those open-ended questions that require detailed explanations.
According to Reddit experts, breaking down complex questions into smaller, related queries encourages the AI to provide focused and detailed responses.
Suppose you’re struggling to come up with a marketing plan, and you prompt AI with, “How do I create a marketing plan?” You’d receive the same unhelpful, rephrased answers with no immediate steps.
Instead, break down the prompt into small fragments to get a detailed, actionable response. See the below prompt for an example.
Prompt: I’m looking to create a marketing plan for a small business selling watches. Please frame your response in the sequence below:
- What is the global market size for premium watches?
- Discuss the best marketing strategy to penetrate the market with real-life examples.
- Based on the answers, outline the steps to create an effective marketing plan for a small-sized business with less than 10 employees.
Chat link: https://chat.openai.com/share/b5c8e622-2369-43fa-a576-a6df23dae1dd
8. Utilizing Priming
Priming is a technique used to configure the AI to respond exactly the way you want it to. It helps you structure your prompts in the right manner. Reddit users suggest two ways for priming the model:
- Set expectations: Specify the desired format, style, or type of response you seek.
- Mention constraints: Include any restrictions/limitations to your response. For example, word limit, business size, budgets, and more.
Prompt: I’m looking for a comprehensive yet concise summary of the latest advancements in renewable energy technologies. Please provide this in a bullet-point format, with each point not exceeding two sentences.
Focus on developments within the last two years. The summary should be suitable for a business audience, particularly small to medium enterprises (SMEs) considering renewable energy solutions.
Limit your response to 300 words to ensure brevity.
Chat link: https://chat.openai.com/share/058c2c81-bf02-4fd0-b4f2-c080f1ea1c16
9. Expand Response Length Limits
Despite being highly sophisticated, there’s a word limit to how long ChatGPT responses can be. The maximum response length of GPT-3.5 is 4,096 tokens (4000 words), and GPT-4 is 8,192 tokens ~ close to 6000 words.
While these limits seem sufficient, considering the AI use cases, they are not. For someone looking to generate an entire book or documentation, the AI response limits are major roadblocks in using the tool at optimum levels.
However, there’s a hack to extend the response length limits.