The Limits of Automated Machine Learning

Every week or every two weeks we see another tool. What comes next and where's the limit? I think I have to stop predicting what is coming next because it's always, okay, I think the code in the printer was not that, let's say, hidden to some guys. But I wasn't aware. Maybe I was not too carefully listening. So it's really difficult to see what comes next. Of course, this will improve clearly. And Mark, what comes next?