Copilot is a service that was launched by Microsoft-owned GitHub and OpenAI in June of 2018. It offers suggestions for entire lines of code within development environments such as Microsoft Visual Studio.
Copilot is an artificial intelligence (AI) model called Codex that was trained on billions of lines of public code to suggest additional lines of code and functions given the context of existing code.
Copilot is available as an extension that can be downloaded, and it is powered by the model.
Copilot can also suggest a method or solution in response to a developer’s description of what they want to achieve (for example, “Say hello world”) by drawing on its knowledge base and the context in which it is currently being used.
Prior to this, Copilot could only be accessed through a technical preview. However, after indicating that the tool would become generally available this summer, GitHub announced today that Copilot is now available to all developers.
This comes after the company indicated earlier that the tool would reach generally available status this summer.
As was mentioned earlier, it will be free for students as well as “verified” open source contributors; initially, this will include approximately 60,000 developers chosen from the community as well as students participating in the GitHub Education program.
During the preview period, GitHub reports that there were 1.2 million people who signed up. According to the company, the percentage of newly written code that Copilot is suggesting has increased to 40 percent, up from 35 percent earlier this year.
According to statements made by Ryan J. Salva, Vice President of Product at GitHub, in an email to TechCrunch, “Over the past year, we’ve continued to iterate and test workflows to help drive the “magic” of Copilot.”
“We used the preview to not only learn how people use GitHub Copilot but also to scale the service in a safe manner.”
The changes that developers make cause Copilot to adapt to those changes, and it learns to match particular coding styles so that it can autofill boilerplate or repetitive code patterns and recommend unit tests that match implementation code.
In addition to Visual Studio Code, the Copilot extension library is also available for use with Noevim and JetBrains, as well as in the cloud via GitHub Codespaces.
The release of Copilot to the general public coincided with the addition of a new feature called Copilot Explain. This feature translates code into descriptions written in natural language.
Although it is being presented as a research project, the objective is to provide assistance to novice developers as well as those who are attempting to work with an unfamiliar codebase.
“While it’s obvious that Copilot helps developers finish tasks more quickly, we’re continuing to investigate updates that go beyond that,” said Salva.
“These updates will help developers stay in the flow, concentrate on work that’s more satisfying, and conserve mental energy even as they save time.”
It is important to share preliminary findings from the research that we are currently carrying out because they can serve as an illustration of the effects that we have seen.
As part of the experiment, we are requesting that developers write an HTTP server, with half of them utilizing Copilot and the other half not.
According to the preliminary data, when using Copilot, developers are not only more likely to finish their tasks, but they also complete them in roughly half the time.
Because of the intricate nature of AI models, Copilot is still not a perfect system. GitHub has stated that it is in the process of developing a filter that will help detect and suppress code that is repeatedly repeated from public repositories.
Additionally, it has stated that it has implemented filters to block emails when they are displayed in standard formats as well as offensive words.
However, the company admits that there is a possibility that Copilot could generate insecure coding patterns, bugs, and references to obsolete APIs, as well as idioms that reflect the less-than-perfect code present in its training data.
Salva continued by saying, “This is just the beginning of AI-powered development tools, so it will be exciting to see how developers use Copilot over the next few months and years from now—and in tandem, how we advance the product.”