If and until the copyright issues around the generated code have been worked out, I would strongly advice against using these tools to generate code for open source projects.
For those that are unaware, the model used to generate code was trained on publicly available open source projects. Many (and I'd guess most) of these projects carry licenses that require attribution or even the whole license to be included when their code is reused. Copilot has been shown to reproduce such code without the required legal additions. Therefore any use of Copilot generated code is opening your project up to the possible risk of copyright infringement and, worse, the insult that is reusing open source code without giving the original author the credit they are due.
It's an interesting experiment, but as a tool for work it isn't yet fit for purpose and could actually harm the user, their project, and the open source community as a whole.
If and until the copyright issues around the generated code have been worked out, I would strongly advice against using these tools to generate code for open source projects.
For those that are unaware, the model used to generate code was trained on publicly available open source projects. Many (and I'd guess most) of these projects carry licenses that require attribution or even the whole license to be included when their code is reused. Copilot has been shown to reproduce such code without the required legal additions. Therefore any use of Copilot generated code is opening your project up to the possible risk of copyright infringement and, worse, the insult that is reusing open source code without giving the original author the credit they are due.
It's an interesting experiment, but as a tool for work it isn't yet fit for purpose and could actually harm the user, their project, and the open source community as a whole.
Thank you for sharing this 😉