I've been experimenting with it.
All I can really say is that there is some utility in its completions. But I found that you still have to go through everything it comes up with, because there is a lot of cruft out there that it will happily give to you.
As long as you are really precise with your comments and how you structure your code and comment prompts, it seems to do really well. Almost like magic. But it easily goes off on tangents, quite like how you can get your cell phone text messenger to produce crazy sentences by prompting it with a word and accepting completions as-is.
Also, it is all too easy to send it into a loop where it continually repeats lines or complete code sections it has already produced.
All the stuff about it going against licenses... I don't know if that is actually the case, BUT it is important to note that Github has standardized licenses for the repos, and therefore there is absolutely no excuse if co-pilot isn't obeying them. They're part of Github's work flow!
Until I see something that I know came attached to a license that wouldn't allow it, I'm not falling for that particular claim. I would expect them to obey the individual licenses that Github provides each repo.
Is it worth it? Not yet. But I have no doubt it will be. I have another month to try it free. But I have a feeling I'll just let it slide as a novelty item I once worked with. I see its potential, and I think it is an amazing toy. But if I was developing co-pilot, I would handle it considerably differently.