This is an interesting thought. But a big concern is one that is already an issue in the corporate world: the biases of the AI “judges” employed to make determinations regarding customers. I'd fear the training materials could result in anything but a fair arbiter—and, moreover, one that is blind to its own biases and those of the precedent that trained it.

it is hard to imagine a professional role that could be more easily delegated to artificial intelligence than that of a civil trial judge

www.americanbar.org/...