One thing that humans will continue to be better at than machines for a very long time is creating interesting stories for other humans. There is no machine that has nearly written a Faulkner novel, or even written up anything more complex than a baseball game (which is itself impressive, though the template is not much more complex than a complicated madlib). When you read machine written stuff, it stinks of the machine that made it. The wriber software is little different at this stage, as evidenced by some of the mangled "ideas" it's generated for others on this thread.
Software and data absolutely have a place in any good editorial strategy, but presuming an algorithm can furnish ideas that will perform better than humans armed with data-driven insights is a mistake in my opinion.
Our content strategists are armed with boatloads of social data, they identify trending topics, top performing articles from competitors etc., and all of that informs the human decisions they make as the editors of their brand publications about what gets written.
At the end of the day, though, I'm going to leave the decision making (and a good part of the ideation) to the talented humans that make up the L&T Co. team.
I'm more compelled by software that helps my team come to data-driven insights themselves than by software that purports to "do the hard part" for them.