AI Assistant using GPT-4/AutoGPT with initializaiton sentence.

The most practical way I can see is to use AutoGPT or OpenAI’s API. My idea is to send an article explaining what the Needpedia is and how to use it to the model, then send users’ response to the model and response back its reply. This solution has been proved possible via an experiment (on GPT-4 model), which the conversation will be attached below, and should qualify the lvl1 function (regarding to the post). Here is the initialization sentence: .

Take note of the first paragraph of the sentence since it will explain some details about the rules that we should make in order to keep the model and users safe.

One small thing I must mention, the OpenAI’s API does not support continuous conversation naturally, however, it does allow you to add multiple messages into one single request to achieve a similar functionality.

The real problem of it, however, is that the cost will increase rapidly, which can be demonstrated via this lua code: f = function(x) if x <= 0 then return 0 else return f(x-1) + x end end or this mathematic function: f(x) = (x+1)x / 2. Therefore, it is important to limit users’ access frequency.

The following is the conversation (init. sentence omitted): .