Initial system instruction

To set up initial system instruction on TypingMind (a.k.a system message on ChatGPT):
  • Go to Model Settings
    • Image without caption
  • Navigate the Initial System Instruction section
  • Input your instruction to guide the chat model behavior during the conversation.
    • Image without caption

What is TypingMind’s initial system instruction?

TypingMind’s initial system instruction is the first message provided by you to the model in order to give the AI assistant a general rule or guideline to follow in the conversation. This could include instructions about the tone, style, or context in which the model should respond.
For example, if the system instruction is to "Write in a style of Shakespeare", ChatGPT would provide responses in the style of this greatest writer for any of your inquiries.
Image without caption
The initial system instruction plays a crucial role in shaping the AI's responses and the overall flow of the chat.

Formula to craft a good ChatGPT system instruction

While there is no one-size-fits-all formula, here are some guidelines to help you create an effective instruction:
  1. Be clear about your purpose:Β clearly define the desired purpose and the intended outcomes for ChatGPT.
  1. Provide context:Β offer specific and relevant background information to help ChatGPT understand the context
  1. Specify format:Β tell the assistant which output format you want. It can be a step-by-step, a table format, a calendar or more.
  1. Set tone: define the desired tone, whether it's formal, casual, instructional, or inventive.
  1. Address the potential limitations:Β keep in mind the limitations of the model, such as avoiding sensitive topics or providing disclaimers when necessary. Or ask ChatGPT provide alternative options when you don’t feel its output a fit to you.
Image without caption
Please note that the effectiveness of this instruction may depend on different factors. You don’t need to cover all of 5 above best practices in your initial instruction, it's essential to iterate many experiments and adapt the instruction to suit your goals.