Harnessing GitHub Copilot as your Coding GPS

GitHub Copilot is a coding assistant powered by Artificial Intelligence (AI), which can run in various environments and help you be more efficient in your daily coding tasks. In this new series of content, we will show you how GitHub Copilot works in Visual Studio specifically and how it helps you being more productive.

 

Time for another short video! Today Bruno Capuano talks about using GitHub Copilot to help with your code, but without "letting the Copilot fly the plane" (yes, we know that in real life, copilots sometimes actually fly the plane!! ). 

 

 

Not a replacement

What Bruno means is that GitHub Copilot is there to assist you with your coding activities, but it's not there to replace you. One of the core principles of Artificial Intelligence at Microsoft is that it should be there to help humanity, not replace humanity. As Satya Nadella once said, AI must be designed to maximize efficiencies but without destroying the dignity of people. 

 

This is why it's crucial to always validate the output of Large Language Models (LLMs) and make sure that it is correct. Sometimes, LLMs don't do exactly what you want. This phenomenon is known in the industry as hallucinations. Sometimes, Copilot will propose code that actually works, but doesn't do what you wanted. Or sometimes, it will even propose code that doesn't compile. Nonetheless, in average, Copilot "gets it right" more often than not. And even when it doesn't, it can guide you on the right path.

 

A new UI paradigm

LLMs are a new way to interact with computers. Instead of inputting data by pressing buttons and getting results computed by an algorithm, you use natural language, like you would with another human. In fact, you can now already interact with some LLMs with images or videos as input. The results are not determined by an algorithm but by a complex process involving probabilities and the model's training.

 

 

Yes, your job may be changing...

One thing the user needs to get familiar with is that inputting the same request multiple times may very well produce a different result every time. As the usage of LLMs is growing in tech and in other industries, our jobs are changing. Now is a great time to learn how to use GitHub Copilot and other AI-powered tools to add these skills to your toolbox. We hope that this series of videos helps you to keep learning.

 

More information

As always, you can find more information, in our collection with resources here. Also, you can see the full-length video here.

 

Stay tuned to this blog for more content posted regularly. And of course, you can also subscribe to our YouTube channel to get more video content!