Taming ChatGPT

Taming ChatGPT
0
222
6min.

Tame ChatGPT. How to formulate queries for better results

Taming ChatGPT

The ChatGPT neural network is one of the most advanced neural systems with a huge database and self-learning. Today, more than 200 million people use ChatGPT, an artificial intelligence-based chatbot from OpenAI, every week. But what is the best way to interact with ChatGPT? There’s really no right or wrong way to communicate with this machine learning-based tool. You can really write anything to it and get certain results. However, there are some guidelines that will help you formulate the right queries to get the best results. The developers of the language model, OpenAI, are constantly making adjustments, so ChatGPT is developing and updating constantly.

How to use ChatGPT effectively

It is important to understand that ChatGPT is not a search engine like Google, but a real assistant. It knows a lot of things and speaks several languages. But it is very important to give it the right query – a prompt. This way, it can better understand its task and provide the most accurate answer.

Taming ChatGPT

So, the correct query algorithm is:

  1. It is best to start the query with action words, for example, “create”, “invent”, “invent”, etc. You can specify the style of the email, define the role of the author, and specify the format of the material. It is also useful to say a few words about the topic to give the neural network a clearer idea of the context. Add more context and specifics, with them the answer will be much more accurate.
  2. Avoid abstract concepts and negations. It is difficult to predict how a neural network interprets such concepts as “friendship”, “beauty”, “what will happen in a thousand years?”, “or what is love”. Instead of “love”, you can write “a young couple on a date at an amusement park” to get a more specific result.
  3. Add a character limit to your response, so you set a certain volume. This is especially helpful when writing headlines or a couple of paragraphs.
  4. Give sources for inspiration. AIknows a lot of artistic styles, writers, artists, photographers, and other creative personalities. By including the name of a famous cultural figure in the query, you can get an interesting and unique result.

It is important to remember that when preparing images or texts for commercial purposes, such requests should be avoided, as such experiments can lead to possible copyright infringement claims.

  1. Set examples. You may want to prepare the same content for different channels – Twitter, Facebook, or LinkedIn. The text should be adapted to the tone and format of the target channel.

Example for Twitter: Compile the generated text into several tweets. Remember that the maximum length of one tweet is 280 characters. Use short sentences and don’t stretch them over several tweets. Start your tweets with consecutive numbering.
Example for Linkedin: Please format the generated text as a post on LinkedIn. Keep in mind that the maximum length is 3000 characters. Structure the main points of the text in a bulleted list. Start with a captivating teaser sentence and end with a call to action for greater engagement.

Life hack. If you doubt whether you can clearly formulate a request to ChatGPT, ask it to do it for you. Give the chatbot a task: “Write a prompt for me that would more accurately describe the following topics: “How has the gambling industry in Ukraine changed over the past 2 years”. Then copy and paste the query into the chat.”

When forming a question to ChatGPT neural network, you should avoid:

  • questions that do not have a clear answer
  • open questions or ambiguous wording

Taming ChatGPT

A few more life hacks for creating high-quality instructions for AI

Pay attention to talkativeness.A chatbot can give both a long answer and concise hints to the prompt you created. If most of your chat tasks require similarly sized answers, you can limit the conditional “chat verbosity” in the same field – Verbosity. Let’s look at 5 Verbosity gradations that you can use to control the detail of your answers.

  • Level 0: Very concise answers, i.e. factual or direct instructions without further explanation
  • Level 1-2: Answers will be more detailed, with some explanation or context
  • Level 3-4:These answers will be more extended, with examples, explanations, and deeper analysis.
  • Level 5: The most detailed and informative answers that cover all aspects of the question with extensive explanations and possible solutions.

Conclusion

Prompts/requests/instructions are a key tool for interacting with the ChatGPT intelligent model. With their help, you can control the text generation process and get more accurate and high-quality answers. However, it is worth remembering that you should be very careful with ChatGPT when it comes to updates and new data, do not rely on it completely and always double-check what you write. It can easily share outdated documentation or generate an old version that won’t work.

Share your thoughts!

TOP