3 ms·
I think the bottom line is that Microsoft’s (and thus other for-profit AI initiatives) stance is that any and all data is fair game regardless of license or aut
by comfypotato 3y ago
I think the bottom line is that Microsoft’s (and thus other for-profit AI initiatives) stance is that any and all data is fair game regardless of license or authorization. This results, in their opinion, from the fact that the AI alters the data, changes the output, and is otherwise “inspired” by the data in the same way an artist might be inspired by another without copyright infringement.
- nologic01 3y agoThis sounds very dodgy. Will somebody be checking the degree of such "data alteration" and verify that the "AI" is actually inspired rather than copying? To me this feels like its opening up the door for the elimination of copyright as any algorithmic layer interjected between scrapped data and end users could claim to be "inspired".
- comfypotato 3y agoWelcome to the discussion lol. People have already provided examples of copilot producing niche code verbatim thereby proving their intuition incorrect. It’s a whole mess that will take years to be cleaned up by new legal conventions.
- Jevon23 3y agoIf my compiler was “inspired” by leaked Windows source code and altered it into a new form then I think their opinions on the matter would be very different.
- comfypotato 3y agoNot a great example: if Apple’s code leaked, theoretically they wouldn’t include it in the training as it’s not supposed to be seen by the public. If it’s public you can be inspired by it (so their logic goes).