Microsoft made a coding AI based on your code, yet software engineers seems to be happy having such bots. So software engineers seems to have a different mindset about these things.
The coding bot can be used by me to make more money / work less. I dont think WAME are giving the AI voice acting bot to the original voice actor so he can put it to work making money are they ?
There are a lot of nuances based on the type of license associated with the code that was used to train copilot.
Have there been breaches of license agreements? I think so. I have no way to prove it (there may be organizations who have). Should those breaches be litigated? Yes.
However, in that same vein, this should be litigated as well.
I would love to be able to generate an open source application by using others open source code! Would be amazing if it could just collect all sources in a similar way to how one would make references/citations in an essay, and “forcing” the code to be released as open source as well. Basically build on all the open source snippets that people release to power the future.
My code there is all MIT licensed, I think it's totally fair to interpret permissive licenses as allowing AI model training. They don't train on private repos.
For people who work on GPL'd stuff sure, it's more questionable.
MIT licensed work still requires attribution, which (as far as I know) current AI models / training practices are unable to handle properly. From the MIT license text:
> The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
Copilot's unlicensed handling of people's code is one of the few areas where "you can't just ignore the license when it comes to training/reproduction" is popular around here.
