Users may now personalize their interactions with OpenAI’s AI-powered chatbot, ChatGPT, in a new way.
On Thursday, a few X users complained that ChatGPT’s online custom instructions menu had been updated with more fields. Users may now choose “traits” they would want the chatbot to have, along with additional information they would like ChatGPT to know about them, their job, and a chosen moniker or nickname. OpenAI recommends characteristics such as “Gen Z,” “Chatty,” and “Encouraging.”
Note that this has nothing to do with ChatGPT’s memory capability, which allows users to specifically instruct ChatGPT to remember—or forget—something. Confusingly, ChatGPT “memories” might include personal information as well as a wider variety of information that is pertinent to the conversation.
For this reporter, the alternatives have not yet shown. TechCrunch has contacted OpenAI for comment; if we hear back, we’ll update this post.
The new menu is probably more of a cosmetic update than a significant technological advancement. The style and tone of ChatGPT were determined by quick engineering, which was employed in the previous personalized instructions feature. In essence, it “primed” the ChatGPT AI with a paragraph of user-supplied instructions so that its responses complied with those specifications and preferences.
As far as we can determine, prompt engineering is still used in the new menu. things just delivers things in a more approachable manner.
As previously stated, OpenAI moderates custom instructions to ensure they comply with its conditions of use. Hopefully, the updated customization options haven’t affected that.
As its user base expands, OpenAI has worked to enhance ChatGPT by incorporating features like live web search and a “Canvas” interface designed specifically for writing and coding tasks. More than 300 million individuals use the chatbot every week, according to the company’s December report.