As a data scientist, I have the lingering fears of job automation. At least for applications-oriented work (as opposed to research). There's no reason Auto ML software couldn't take away most of the difficulty of model building in simpler scenarios. I mean, the software is already here. Just not widely used. And as computing power increases, no reason they can't tackle more of the complex problems where you might need to call in an expert. Feature generation based on the context of the industry, etc. Translation of findings into business-understandable language. One new algorithm or software package can automate much of what a data scientist might do. I mean, the unofficial slogan for some of the data science positions I've found were to "automate myself out of a job."

That's why, although I LOVE what I do, part of me thinks I need to make sure I'm in a position to move into management at some point - because I see a lot less ability for automation in areas that require "people" skills over the next few decades. On the other hand, I LOVE what I do. What's that French phrase? C'est la vie? I think it translates into "you're screwed, one way or another."