Github Copilot training and escape room mysteries

Artificial intelligence’s capabilities and assistance have expanded into every domain, and software development is no exception. On our training day, we introduced a new tool that promises to revolutionize coding: GitHub Copilot. Spoiler; coding is under revolution! 

GitHub Copilot training 

GitHub Copilot can be described as an intelligent AI pair programmer. While you code, it sits on your shoulder, performing manual tasks, answering programming-related questions, and enhancing the quality of your work with suggestions that can make your code more efficient. 

It was born through a collaboration between GitHub, OpenAI, and Microsoft. Copilot was trained on billions of code lines available on GitHub and uses OpenAI’s Codex AI model. It is integrated into development environments like Visual Studio. 

Copilot closely follows your work in the development environment. For example, it can create a new class with its properties based on the description you provide. 

We’ll create a method with its assistance, which will request user information and “mock” data for testing the code. 

You can interact with Copilot in various ways. You can open a chat window in the corner of your development environment to request suggestions for improving your code. 

Let’s implement the suggestion for a more efficient way to gather user information into a collection. 

Let’s request a suggestion for the unit test of the method, and with a little fine-tuning, unit tests will be generated as if by magic. 

What crowns Copilot’s intelligence is its understanding of the context of your code. It keeps a record of your interactions and continuously learns from your accepted suggestions. It assists in commenting and documenting your code. 

One of the significant advantages of Copilot is how much mundane manual work it eliminates for developers, saving time and reducing errors. Coders also benefit from learning new techniques as Copilot justifies its suggestions and explains concepts. 

It’s crucial to note that the quality of Copilot’s suggestions depends on the programming language used. The more code lines of that language and the type of code available in the training data, the smarter Copilot becomes in that language. The most common languages in the dataset are Python, Java, and JavaScript. 

The coder’s responsibility for code security and quality should be emphasized. Copilot relies on the content of your files and the history it has collected while working with you. However, you can prevent your data from being used in general product development. Your data remains encrypted during transmission and in the REST API. Access to your data is limited, and it won’t be used as suggestions for other users. 

Recreation

After the training, Hotel Sveitsi fulfilled all our relaxation wishes. Escape rooms, the autumn sun on outdoor trails, sauna, hot tub, snacks, and cozy spaces for socializing. 

We successfully escaped from escape rooms without AI, but it has become an invaluable support in software development. Contact us to continue the conversation about how you can leverage AI in your everyday applications.