GitHub Copilot Is Now Live: First Look

That week, Github launched its AI programming tool called GitHub Copilot to the masses. For a whole year, while the service was being tested, the beta version was available only to a limited circle of developers. And now it is finally live for all individual developers having a GitHub account.

The name and idea of the service are supported both by the logo with the head of a robot in a pilot's helmet and the slogan - "Don't fly solo".

GitHub Copilot logo.png

As officially stated, Copilot is supposed to “help you write code faster and with less work”. Sounds great! Still, not everyone is satisfied with the service. Let's take a look at the main  features to figure it out.

In this article:

How does GitHub Copilot work?

The tool is programmed to suggest ready-made lines of code. You just start writing, and Copilot takes words right out of your mouth, or rather from the keyboard, and finishes the block for you. Another way to get a suggestion is to write a comment describing the required logic.

The tool is based on the generative pre-trained language model Codex which helps Copilot extract context from comments and code. There are over a billion ready-made lines and dozens of languages, including human ones, in its database. It's not hard to guess that they come from open source projects submitted to GitHub.

As a suggestion, you can see not only single lines but also entire functions. Once the recommendation appears, you can either accept it, reject it, or edit it.

GitHub Copilot.png

Github Copillot is available as an extension for Visual Studio Code, Visual Studio, Neovim, and the JetBrains suite of IDEs.

What’s the price of GitHub Copilot?

The tool is currently offering a 60-day trial so you can test it for free. And this is a pretty good period to properly “try out” the tool and understand if you need it at all.

After these two months, you can continue using the $100/year plan (extended automatically), switch to the $10/month one, or cancel your subscription altogether.

What are the cons of GitHub Copilot?

The developers of GitHub Copilot themselves do not deny that the tool does not write perfect code. They even state that “the code may not always work, or even make sense’.  According to their research, only 26% of recommendations are accepted by users.

Here are some shortcomings that can be identified even before you start using the service, only by carefully reading the FAQ section:

  • The tool doesn’t test the code it suggests.
  • Suggested code may contain bugs, insecure coding patterns, and verbatim from public code. 
  • You are solely responsible for the code written in conjunction with Copilot.
  • Comments written in languages other than English may be handled less efficiently.
  • Some programming languages may perform better than others.

It follows that you must be extremely careful with the code you run to avoid compilations. Using the service only somehow speeds up the coding process, but does not relieve you of the responsibility for checking the code.

By the way, using the Notify.Events service, you can easily set up sending messages from many developer services to IMs or other tools of your choice. For example, in Signal, Viber, Discord, Slack and other 40+ apps. And then let your colleagues do the same!

Notify.Events - DevOps sources directory.png

Learn how to create your own thematic channel and start receiving notifications in a convenient way today.

Margarita Ramsten | Jul 1, 2022 Share it: