GitHub's Copilot feature is an Artificial Intelligence model developed by OpenAI called Codex

GitHub Copilot's voice feature will allow programmers to code using their voice. GitHub's Copilot was introduced last year and uses Artificial Intelligence to help speed up mundane coding tasks by suggesting how to complete lines. According to GitHub, its new GitHub voice assistant can understand natural language requests for Copilot to suggest a code snippet, or summarize what a specific section of code does. But even if a developer doesn't want any code suggestions, it can serve other practical use cases such as helping them navigate a codebase by saying something like "Hey GitHub, go to line 34," or even control the IDE by toggling to zen mode.

Behind GitHub's Copilot feature is an Artificial Intelligence model developed by OpenAI called Codex. OpenAI claims that Codex has a broad knowledge of how people use code and is "significantly more capable than GPT-3" in generating code.

Microsoft-owned GitHub announced at its annual conference this week an experimental feature that pairs Copilot with natural language recognition. The result is that developers can request code snippets, or summarize what a section of code does, using just their voice.

Aside from writing and editing codes, 'Hey, GitHub!' will allow programmers to do code navigation by simply indicating the line where they want to view (e.g., "Hey, GitHub! go to line 34," "Hey, GitHub! go to method X," or "Hey, GitHub! go to the next block,"). It also accepts other VisualStudio Code commands such as "Toggle zen mode" and "run the program." Additionally, experiment participants will be able to have quick access to code summarization in Hey, GitHub!, giving them summary explanations of certain code functions.

GitHub seems to think it will be a worthwhile investment, pointing to an internal study that shows how Copilot has been able to increase developer productivity by 55% with its coding suggestions. That's not the only benefit, though, as the same study showed that Copilot also has the effect of boosting developer happiness. Not only do developers get a productivity boost, but they can also focus more on business logic and discover ideas that they might not have considered off their own backs.

Microsoft and OpenAI have found themselves in hot water over Copilot. A class-action lawsuit was launched against the companies this week to determine whether the feature infringes on the rights of developers by scraping their code and not providing due attribution.

"Copilot leaves copyleft compliance as an exercise for the user. Users likely face growing liability that only increases as Copilot improves," wrote Bradley M. Kuhn of Software Freedom Conservancy earlier this year.

"Users currently have no methods besides serendipity and educated guesses to know whether Copilot's output is copyrighted by someone else."

While GitHub's Copilot is currently only useful for completing lines of code – similar to Gmail's Smart Compose – Microsoft has touted that the feature will improve to suggest much larger blocks, like the entire body of a function.

Generative AI can be enormously beneficial for programming, as evidenced by a recent survey of more than 2,000 programmers finding that Copilot users are likely to code faster and feel "more fulfilled with their job" Nearly three-quarters said Copilot enabled them to "focus on more satisfying work," because the AI helped dispatch the more tedious and repetitive elements of coding. The voice connection will likely improve those numbers, especially for those who may prefer talking through their code without typing.