Artificial intelligences are increasingly present in our day to day. More and more jobs are using this type of algorithm to facilitate the work of the workers themselves, sometimes even being able to replace them. Although it is true that there is still a long way to go to get an AI to think and act like us, these are becoming more complete and have increasingly advanced self-learning functions. And we’ve reached a point we thought we weren’t going to be able to get: teaching an AI to program.
GitHub is not only a platform where you can create all kinds of open source projects, but it is also a huge source of code, accessible to any interested user. Taking advantage of this source, OpenAI and Microsoft have been collaborating together on a project capable of programming on its own: GitHub Copilot. After a long time of teaching and learning, AI is finally ready for users of the Microsoft IDE, Visual Studio Code, to use in their projects.
This is how GitHub Copilot works, the AI capable of programming
The programming IDEs already had some smart features, like syntax highlighting, or some auto-complete feature. But nothing is as complete as what this new Artificial Intelligence offers us. Visual Studio Code users can now download the new extension, called GitHub Copilot , to start programming with it.
This extension will allow us to auto-complete functions, of course. But its possibilities go much further. The new wizard is able to understand the code we are writing and will be able to synthesize code to complete functions automatically. The possibilities that Copilot offers us are:
- Convert comments to code . We can write what we want to do in a logical comment, and Copilot will generate the code automatically.
- Auto-fill repetitive code . Simply create a couple of inputs for a repeating function and the AI will do the rest.
- It allows us to generate test code . So we can see which one best suits our development.
- Show alternatives . If we want to give our program a different approach, thanks to Copilot we will be able to see different examples and choose the one we want.
This AI still works only in natural English (for example, to convert comments to code). Thanks to OpenAI Codex and analysis of all public GitHub repositories, this AI has achieved success rates of over 60%. In addition, as we use it, each time it learns more to generate the codes that best suit our needs.
Get started with GitHub Copilot now
At the moment, Copilot is only available to a very limited group of users. This is because, in order to use it, it is necessary to have state-of-the-art hardware, compatible with Artificial Intelligence and machine learning functions. Microsoft and GitHub offer beta access to a limited number of people for free . When As time passes, the number of users will increase until, finally, it will be available to everyone. We can join the waiting list to receive the preview at this link .
The preview of this AI is free. However, if it is successful, they plan to launch a paid version of Copilot, although it is not known what features this one will have. And for now, it ‘s exclusive to Visual Studio Code .
As for telemetry, one of the most feared aspects by users, Microsoft tries to call for calm. It is true that to work, it is necessary to constantly send code to the GitHub servers, but this data is transmitted, and stored, safely. All code is analyzed and processed by a machine , and only authorized personnel could access that data, should it be necessary (for example, to detect abuse). In the future, Microsoft plans to allow users to control this telemetry, but for now, it is mandatory.
