Not long ago we share here on the blog the news of «Copilot» which is an AI that promises to save time by helping with user code writing with their own suggestions that are based on billions of lines of public code whose users have publicly contributed to GitHub, using an artificial intelligence system called Codex from research firm OpenAI.
While Copilot is a huge time saver and GitHub describes Copilot as the AI equivalent of "pair programming", in which two developers work together on a single computer. The idea is that one developer can come up with new ideas or spot issues that the other developer might have missed, even if it requires more hours of work.
In practiceHowever, Copilot is more of a time-saving utility tool as it integrates resources that developers would have to look elsewhere. As users enter data into Copilot, the tool suggests snippets for them to add with the click of a button. That way they don't have to waste time looking for API documentation or looking for sample code on sites like StackOverflow.
A neural network on GitHub Copilot is trained using massive amounts of data, made up of code - millions of lines uploaded by the 65 million users of GitHub, the world's largest platform for developers to collaborate and share their work.
The goal is for Copilot to learn enough about code patterns to be able to hack on its own. You can take the incomplete code from a human partner and finish the job by adding the missing parts. In most cases, it seems to be successful in doing this. GitHub plans to sell access to the tool to developers.
Like most artificial intelligence tools, GitHub also wants Copilot to get smarter over time based on the data it collects of the users.
When users accept or reject Copilot's suggestions, their machine learning model will use that feedback to improve future suggestions, so the tool can become more human as it learns.
Shortly after the launch of Copilot, some developers began to be alarmed by the use of public code to train the artificial intelligence of the tool. One concern is that If Copilot reproduces large enough chunks of existing code, it could be copyright infringement or launder open source code for commercial use without the proper license.
About it The Free Software Foundation has announced that it has launched a call funded to request technical reports on the implications of Copilot for the free software community
“We already know that Copilot as it is is unacceptable and unfair, from our point of view. It requires running non-free software (Visual Studio or parts of Visual Studio Code) and Copilot is a service as a replacement for the software.
The reason is that Copilot requires running non-free software, as Microsoft's Visual Studio IDE or the publisher of Visual Studio Code, maintains the FSF, and constitutes a "service as a substitute for software," which means that it is a way of gaining power over other people's computing.
Since as such Copilot is a Visual Studio Code extension that uses machine learning trained in freely licensed open source software to suggest lines of code or features to developers as they write software.
However, Copilot raises many other questions that require further consideration.
“The Free Software Foundation has received many inquiries about our position on these issues. We can see that Copilot's use of open source software has many implications for a large part of the free software community. The developers want to know if training a neural network in their software can really be considered fair use. Other people who might be interested in using Copilot are wondering if code snippets and other material copied from repositories hosted on GitHub could result in copyright infringement.