Chat Bot using Chat GPT 3.5 in C#

Unleashing the Power of OpenAI and C# to Build Your Own Chatbot

This code employs the power of OpenAI’s chat API to create a sophisticated chatbot. The bot’s neural network is trained on a vast corpus of text, enabling it to respond to user input with impressive fluency and flexibility. The chatbot is initialized with a set of messages that introduce it to the user, and then it waits for the user’s input.

Once the user enters a message, the code sends that message, along with all previous messages, to the OpenAI API for processing. The API uses cutting-edge natural language processing algorithms to generate a response, which is then parsed and displayed to the user. This process repeats indefinitely, allowing for a fluid and engaging conversation between the user and the bot.

The code uses a combination of asynchronous HTTP requests and JSON parsing to communicate with the OpenAI API and handle the responses. It also incorporates Spectre.Console, a powerful library for styling and formatting console output, to display the messages and responses in an attractive and readable format. Overall, this code represents a sophisticated and powerful example of how modern AI technologies can be leveraged to create compelling conversational interfaces.

Simply create a new c# .net console application and paste the following code into program.cs

// See: https://platform.openai.com/docs/guides/chat/introduction
// See: https://platform.openai.com/docs/api-reference/chat

using Newtonsoft.Json;
using Spectre.Console;

const string key = "YOUR API KEY";
const string url = "https://api.openai.com/v1/chat/completions";

// Initialise the chat by describing the assistant,
// and providing the assistants first question to the user
var messages = new List<dynamic>
{
    new {role = "system",
        content = "You are ChatGPT, a large language " +
                                    "model trained by OpenAI. " +
                                    "Answer as concisely as possible.  " +
                                    "Make a joke every few lines just to spice things up."},
    new {role = "assistant", 
        content = "How can I help you?"}
};
AnsiConsole.MarkupLine($"[purple]MACHINE:[/] [blue]{Markup.Escape(messages[1].content)}[/]");

while (true)
{
    // Capture the users messages and add to
    // messages list for submitting to the chat API
    var userMessage = AnsiConsole.Ask<string>($"[purple]USER:[/] ");
    messages.Add(new { role = "user", content = userMessage });

    // Create the request for the API sending the
    // latest collection of chat messages
    var request = new
    {
        messages,
        model = "gpt-3.5-turbo",
        max_tokens = 300,
    };

    // Send the request and capture the response
    var httpClient = new HttpClient();
    httpClient.DefaultRequestHeaders.Add("Authorization", $"Bearer {key}");
    var requestJson = JsonConvert.SerializeObject(request);
    var requestContent = new StringContent(requestJson, System.Text.Encoding.UTF8, "application/json");
    var httpResponseMessage = await httpClient.PostAsync(url, requestContent);
    var jsonString = await httpResponseMessage.Content.ReadAsStringAsync();
    var responseObject = JsonConvert.DeserializeAnonymousType(jsonString, new
    {
        choices = new [] {new {message=new {role=string.Empty,content=string.Empty}}},
        error=new {message=string.Empty}
    });

   
    if (!string.IsNullOrEmpty(responseObject?.error?.message))  // Check for errors
    {
        AnsiConsole.MarkupLine($"[bold red]{Markup.Escape(responseObject?.error.message)}[/]");
    }
    else  // Add the message object to the message collection
    {
        var messageObject = responseObject?.choices[0].message;
        messages.Add(messageObject);
        AnsiConsole.MarkupLine($"[purple]MACHINE:[/] [blue]{Markup.Escape(messageObject.content)}[/]");
    }
}

Ensure to add the following two nuget packages.

<PackageReference Include="Newtonsoft.Json" Version="13.0.2" />
<PackageReference Include="Spectre.Console" Version="0.46.0" />

Click F5 to run or dotnet run from command line!

Here is a sample conversation (ignore my bad spelling, chat gpt does!)

p.s the opening 3 paragraphs was written by the ChatGPT API!