Generative AI isn’t just a buzzword anymore – it’s a powerful force that’s already transforming industries around the world, including software development. The software development industry is slated to be one of the biggest beneficiaries of AI technologies as almost every process therein, from design to maintenance, can be improved with the implementation of AI.
GPT-4, OpenAI’s latest large language model, is already able to proficiently write computer code in most programming languages. Recently, we saw how GPT-4 could write code for a proper 3D game with nothing but short textual inputs from the human end. Demonstrations like these are not only causing programmers to fear losing their jobs, but have also got OpenAI’s CEO Sam Altman himself worried.
Whether AI will take over jobs in the near future is debatable, but there’s no doubt that in its current state, it has the potential to reshape the process of software development entirely.
When ChatGPT first came out, it didn’t take long for programmers to start using it for coding. That is unsurprising considering OpenAI itself obliquely encouraged the use case – the company had demonstrated the chatbot’s debugging abilities right at the top of its press release about its launch.
Soon after, developers started finding ways to plug the chatbot’s programming skills directly into their workflows. For example, the “ChatGPT” extension (by Ali Gençay) for top code editing software VS Code lets programmers interact with ChatGPT from a sidebar conversation window as they coded.
That plugin was ‘unofficial,’ though, and it was only inevitable that one of the software heavyweights would launch an ‘official’ version with a neater integration.
Enter Copilot X by GitHub, developed in partnership with none other than OpenAI. It uses the latter’s latest, most cutting-edge LLM – GPT-4.
Copilot X integrates natively into VS Code and Visual Studio and introduces a chat interface to the editor. GitHub says that it does a lot more than suggest code: “It recognises what code a developer has typed, what error messages are shown, and it’s deeply embedded into the IDE. A developer can get in-depth analysis and explanations of what code blocks are intended to do, generate unit tests, and even get proposed fixes to bugs.”
GitHub Copilot in action (Image: GitHub)
This obviously means that the Copilot will help programmers with a lot more than just missing semicolons in code. GitHub says that it will help reduce boilerplate and manual tasks and make complex work easier across the developer lifecycle. “By doing so, we’re enabling every developer to focus all their creativity on the big picture: building the innovation of tomorrow and accelerating human progress, today,” GitHub adds.
But chatting isn’t the only medium programmers will be able to utilise to interact with the chatbot. GitHub Copilot Chat will also join GitHub Copilot Voice, a voice-to-code AI technology with which developers can verbally give natural language prompts. With technologies like these, who knows, we may ultimately arrive at a time when coders would only need to interact with their PCs to remove the screensaver.
Copilot will also be used for Pull Requests on GitHub, leveraging GPT-4’s power to add AI-powered tags in pull request descriptions. A pull request is an event where a contributor asks a maintainer to review code they want to merge into a project.
Going through scores of documentation pages can be a nightmare, especially for new team members trying to get up to speed. GitHub Copilot is also simplifying the process of browsing those with a chat interface where developers can ask questions about languages, frameworks, and technologies they’re using.
But while the aforementioned features are all new, Copilot itself isn’t. GitHub Copilot was initially released back in 2021 as an AI-powered tool that simply assisted programmers by autocompleting code and saving them time. It leveraged OpenAI’s Codex AI model – a GPT-3 derivative – with its training data including natural language and billions of lines of code picked from GitHub repositories and other sources.
Less than two years since its launch, Copilot is already writing 46% of code and helps developers code up to 55% faster, according to GitHub.
And now, GitHub seems to suggest that Copilot X isn’t just a bunch of features that developers can enjoy. Rather, it’s the company’s “vision for the future of AI-powered software development.”
