In the OpenAI Blog about function calling feature is the following written:
For example, developers can:
- Create chatbots that answer questions by calling external tools (e.g., like ChatGPT Plugins)
What does this actually mean?
Does this mean that, if I create a chatbot locally, that I can make calls to ChatGPT Plugins? If yes, how? And which kind of plugin’s? All or only mine?
Does this also mean, that ChatGPT plugins are not confined to ChatGPT frontend anymore? How to solve the authentications of this plugins then?
Can a custom plugin call other plugins with the function call API?
I hope to get more clarifications about my questions, because this will definitely open up complete new ways (meta-programming for example).
Thank you in advance.
You can now pass an extra set of parameters in the create method of your call to the OpenAI API, called: functions, in that you describe your function and it’s input parameters and function, then when the AI is called it can decide if it needs that function to complete it’s task, if it does it returns to you a json string containing the functions required variable entities and you can then call that function with those variables set, then you can pass that information back to the model as message text and it can process the new information. The model has also been fine tuned to generate json repronounces in this way so your output will now be more consistent.
Thank you for the general answer.
This I already know and how it works from an API standpoint.
To be more precisely, in the context of calling ChatGPT plugin’s with function calls is unclear for me.
Ahh, ChatGPT Plugins and API functions, although similar in function, are not interchangeable, one cannot call the other, it was just a analogy.
Ah, ok. I thought this was a concrete use case as an example.
I will mark your last response as a solution.
Thank you for clarifying.