ChatGPT Plus gets custom instructions so it can remember how you want it to behave

OpenAI has one today potentially impactful function for ChatGPT Plus Subscription Service Users ($20/month): Custom Instructions, a new setting that users can enable when logged into their ChatGPT Plus account. This setting allows the AI ​​chatbot to store information about how the user wants it to react and behave, maintaining this perspective even when the user closes one chat and starts another.

The feature, currently available in beta outside the UK and EU, could be a huge time saver for regular users of the service, as it saves them from having to start with the default ChatGPT interface and then “primeit with the perspective the user wants every time they open a new chat window. In other words, you can type your overarching prompt once and ChatGPT Plus will keep it for as long as you want, even as you give it new requests and questions and close it and start new chat conversations.

Potential use cases

OpenAI cited the hypothetical example of “a teacher creating a lesson plan”. With the custom instruction feature enabled, the teacher “no longer has to repeat that he teaches 3rd grade physics” every time he starts a new chat with the service. Instead ChatGPT Plus will keep this perspective and answer in the future.

Or, if you’re a developer who likes to code in Python, you can store that information in the new custom statement setting, and ChatGPT will return results in Python every time you ask for coding help instead of asking him always have to remember to do this.

Now how to use the custom instructions on ChatGPT Plus

Users can now try the setting on the web or in the ChatGPT iOS app. On the web, it can be accessed by clicking your account username in the lower left corner of the ChatGPT interface, then clicking Settings, Beta Features, and enabling “Custom Instructions”.

Then the user has to close the menu and click on his name again. A new menu option should appear in the pop-up labeled “Custom Instructions.” Clicking it brings up a new screen with two questions that OpenAI asks you to answer in 1500 characters or less.

“What would you like ChatGPT to know about you in order to provide better answers?” and “How do you want ChatPT to respond?”

OpenAI provides “thought starters” for each question to guide your answers, including “where are you located?”, “What do you do for work?”, “What are your hobbies and interests?” for the former, and “How formal or informal should ChatGPT be?”, “How long or short should responses generally be?”, “How would you like to be addressed?” and “Should ChatGPT have an opinion on topics or remain neutral?”

Based on these frameworks, it looks like OpenAI is trying to help users prepare ChatGPT to respond in a custom way for each of them, and keep that customization as long as they have the setting enabled.

While OpenAI currently only allows ChatGPT Plus users to enter one set of custom instructions at a time, the instructions are completely open provided they fit into the 1500 character text boxes. This means that you can actually have ChatGPT respond from multiple perspectives if you enter them in the text box.

The first experiments show potential

VentureBeat experimented with this by typing, “I am a novelist writing a new work of science fiction. Consider each character’s motivations, personalities, and relationships as you build the story,” in the first text box, then provide character descriptions in the second.

The raw results were technically and grammatically sound, though they clearly lacked the unique writing voice and rigor we’d expect from a published novel…for now. And they can presumably be edited by an individual into something that appeals to some readers.

OpenAI’s beta release of its new custom statement feature comes just a few weeks after it released another big new feature, Code Interpreter, which allows users to upload documents, create visualizations based on data they provide, and write ChatGPT code and run in Python.

The new features come at a time of growing opposition to OpenAI and ChatGPT, inclusive lawsuits from authors claiming that OpenAI has scrapped their books in violation of copyright and a similar lawsuit by a famous comedian. There are also complaints about alleged degradation in ChatGPT’s response quality from when the service first became available in November 2022 and when the underlying model was updated to GPT 4 in March 2023. The US Federal Trade Commission is also reported that he was investigating the company about a data breach.

Still, OpenAI is clearly moving forward with what it sees as improvements and useful new capabilities for its signature service and forging new alliances with established names in the media, including The Associated Press and the American Journalism Project.