OpenAI introduces a novel way for users to customize their interactions with ChatGPT, the company’s AI-powered chatbot.
Some users on X on Thursday announced that ChatGPT’s existing online menu of custom instructions has been enhanced with novel fields. Now users can specify their preferred name or nickname, their profession, other things they want ChatGPT to know about them, and the “features” the chatbot should have. OpenAI suggests traits such as “Talkative,” “Encouraging,” and “Gen Z.”
Note that this has nothing to do with ChatGPT’s memory feature, which allows users to explicitly tell ChatGPT to remember or forget something. Somewhat confusingly, ChatGPT “memories” can be personal information – but they can also include a broader range of information relevant to the conversation.
“Introduce yourself to get better, more personalized answers,” reads the message in the refreshed menu.
Options are not yet apparent to this reporter. TechCrunch has reached out to OpenAI for comment and will update this article if it receives a response.
The novel menu is probably more of an aesthetic overhaul than a deep technical upgrade. The aged custom instructions feature used so-called swift engineering to set the style and tone of ChatGPT. It essentially “primed” the ChatGPT AI with a paragraph of user-supplied instructions so that its responses aligned with these requirements and preferences.
As far as we know, the novel menu is still based on rapid design. It just presents it in a more user-friendly way.
OpenAI has previously said it moderates custom instructions to ensure they comply with its terms of apply. This probably hasn’t changed with the improved personalization settings.
As its user base grew, OpenAI sought to improve ChatGPT, adding capabilities such as live web search and a “Canvas” interface tailored for writing and coding projects. In December, the company reported that more than 300 million people apply the chatbot weekly.