the problem isn't even that this technology will eventually replace programmers: the problem is that it produces parts of the training set VERBATIM, sans copyright.
No, I am pretty optimistic that we will quickly come to a solution when we start using this to void all microsoft/github copyright.