OpenAIannounced in its blog that it is deploying a special instructions interface for ChatGPT. This is done to better control the chatbot’s response to prompts. This feature has been available for the Plus plan since July 20 and will be expanded to all users in the coming weeks.
User instructions will allow you to add parameters or requirements that ChatGPT takes into account when creating its responses. The chatbot will memorize this context and work within the instructions in all conversations until the user changes them. This is how OpenAI responds to users’ dissatisfaction with the fact that every conversation with the same purpose had to be started anew.
Now you can specify important context about yourself in the interface: location, profession, hobbies, goals, etc. In another window, you should specify the actual instructions for ChatGPT: the style of the generated response, the role it should play, etc.
Adding instructions will also help improve the plugin experience by providing the right context each time a plugin is called to find or generate the data you need.
ChatGPT will not memorize instructions that violate the Terms of Use, and may also reject instructions that would cause it to violate these policies with its responses.
This feature is still in beta testing, so OpenAI warns that ChatGPT does not always interpret special instructions perfectly – sometimes it can ignore them or use them for other purposes. The company can use your special instructions to improve their products, but you can turn this off through the data controls.
Source: Media Maker
