ChatGPT traditionally struggles with personalization and memory consistency. To fix this, OpenAI introduced Custom Instructions for every user, free or paid.
Anyone can use ChatGPT's custom instruction feature to personalize the responses they get from ChatGPT while making sure the instructions are used across all chat sessions, and there are a few different ways you can use them.
You can use the ChatGPT custom instructions feature to specify safety rules for your ChatGPT account or those used by minors. If you want to ensure your kid is safe when using ChatGPT, you can use custom instructions to generate responses suitable for children under the age of X, where X is your child's age. You can also get more specific if there are certain topics you want ChatGPT to avoid that may not necessarily be inappropriate for that age.
For instance, if you're trying to wean your child off video game addiction, you could ask ChatGPT to avoid providing instructions or tips for video games. The key is being very clear and specific in detailing what types of content should be avoided or limited. Remind ChatGPT to respond helpfully within those defined boundaries. Testing the instructions with sample inputs ensures ChatGPT complies as intended.
Here's a sample prompt you can add to your custom instructions input field.
Please ensure that all responses generated by ChatGPT are appropriate for a child under the age of [10]. Avoid providing any content related to video games or video game instructions. My primary goal is to create a positive and enriching environment for my child, so emphasize promoting positive values and behaviors. Be very strict in adhering to these boundaries. If any input may lead to content that is not suitable for that age or goes against these guidelines, refrain from generating such responses.
Another handy way to use ChatGPT's custom instructions is to provide formatting guidelines for its responses. This could be a key-value pair or a basic formatting template for ChatGPT to follow to help ensure responses are formatted consistently across your chat sessions and eliminate the need for repeated formatting instructions in every prompt. It is also a perfect opportunity to specify your preference in units of measurement, currency, sorting order, date format, or any other formatting details you would like ChatGPT to adhere to.
You'll find this particularly useful if you use ChatGPT within a particular industry or niche where specific formatting rules are used. Here is a great example of how this is used:
Please use the key=>value pair provided to determine how to structure your responses. For instance, when your response relates to history, use the timeline format. history => timeline format, ideas => mindmap format, References => AP, programming steps => flowchart, article => markdown format, measurement => metric units, Names => Last First Middle, Money => USD $0.00, sort => Z-A, Temperature => Fahrenheit
The custom instruction field is a good place to program ChatGPT to respond to certain events in certain ways. Using natural language, you can specify an action you want ChatGPT to perform when a certain event occurs while prompting the AI chatbot.
For instance, in one of our tests, we programmed ChatGPT to pull the data from one of our articles about writing a CV for a tech job anytime we prompt the chatbot to prepare a CV related to a tech role. So, if we were to post a job description for a tech role on ChatGPT and ask it to help draft a resume, it will pull data about the best way to write a resume from that article and ask a series of questions based on that data. Here's a demo of it in action:
Also, if you have a website with your personal or professional information, you could ask ChatGPT to fetch data from that URL when processing a response requiring personal or professional details.
A good example would be a resume. Instead of providing ChatGPT with your professional details whenever you want to write a resume, ChatGPT will automatically pull your professional information and fill in your resume. Of course, it doesn't always have to be a resume; it could be anything, such as a cover letter, email, proposal, etc. Also, it doesn't always have to be a link, as you can get creative with your custom actions.
Here's a sample prompt you can modify to build an interactive custom action that's activated only when a condition is met:
Whenever I reach out for help regarding a resume, I need you to refer to the information provided at "https://www.makeuseof.com/how-to-write-cv-for-tech-job/" to guide our interaction. Based on the insights from that page, you should pose a series of eight targeted questions to me. My responses to these questions will be the foundation upon which you will construct a professional and effective resume tailored to a technology role. This specific procedure should be activated only when I'm seeking assistance with a tech-related resume. Moreover, I would like our exchange to be sequential, where you will present each subsequent question only after I have provided a thorough answer to the one preceding it.
While capable of producing quality writing, ChatGPT's write-ups often lack a consistent personal voice and style. If you use ChatGPT primarily or frequently for creating write-ups and need the chatbot to maintain a consistent style across all sessions, custom instructions might be your way out.
You can provide simple instructions within the custom instructions field to get ChatGPT to adopt your writing style, including pacing, word choices, and tone. How? Start by pasting a copy of one or more write-ups that closely reflect your preferred writing style, and ask ChatGPT to extract a personal preference map. Here's a prompt to help you achieve that:
I want you to extract a Personal Preference Map (PPM) from the write-up I provide in the next prompt. Now, a PPM is a key => value pair of conditions mapped to preferences. Below are examples of key => value pairs:Tone => sarcastic, sassy, and loving. Word choice => formal, complex. Sentence structure => mixed of short and long, mostly short. Explanation style => imagery, vivid, relatable. Reply affirmatively if you understand the task. When I provide the next prompt, extract the PPM using the same logic and formatting used above. The key => value pairs should be separated using "=>" Apart from the tone, word choice, and explanation style, I want you to include 10 other stylistic elements that better capture a writing style. Print the PPM after extraction.
結果應該是這樣的。通讀、複製結果並將其貼上到自訂指令功能的第二個欄位中。
雖然自訂說明可用於實現複雜的目標，但它們的主要目的是了解有關您的更多資訊並根據您的需求客製化響應。您可以定義特定參數，例如您的職業、過敏或特定領域的教育程度和專業知識。這樣，您將收到與您的職業、飲食限制（排除食譜中的過敏原）或教育背景相符的定制回复，以確保每當您尋求信息或指導時，解釋都是根據您的理解水平量身定制的。
以下是一些範例，說明如何使用自訂指令功能來自訂 ChatGPT 的回應，以符合其特定的個人需求和專業背景。
您可以讓ChatGPT 考慮您家人的需求。
這些只是三個簡單的範例，但您可以透過多種方式根據您的需求調整 ChatGPT 自訂指令。 
充分利用自訂指令 預設情況下，無論主題是什麼，ChatGPT 的回應確實令人印象深刻。然而，其平淡、籠統的回應可能並不總是適合您的獨特需求。透過一些創意和努力，您可以利用自訂指令來確保一致、個人化的互動符合他們的獨特需求和偏好。不要滿足於普通的 ChatGPT 體驗。掌控並定制它，以便更好地為您服務。 以上是使用 ChatGPT 自訂指令的 5 種最佳方法的詳細內容。更多資訊請關注PHP中文網其他相關文章！
