The current phase of labor automation seems centered around automation of low intelligence-intensive high-regularity jobs.
These presently include clerks of all manners, secretaries, documentation-related tasks and so forth.
The next phase of labor automation will still be dominated by low intelligence-intensive jobs but with low-regularity semi-creative jobs.
These include intermediate artists and artisans of all manners, including photography, film, and basic design tasks such as web and simpler graphic design. A good example of such automation might be tweening artists in animation industry. Artificial imagery and art. There is indication that current tech is already sufficient for beginner to intermediate level of musical performances. Although that will not be practical or desirable, it shows the tech is there. The possibilities are endless.
These also include various smart engineering assistance and research automation tools, journalism, and other jobs that require a bit of creativity but not much. Such systems will be indispensable for research and news aggregation on the internet. For instance, such a smart scientific survey system will be able to find methods in the vast engineering literature that solve a problem of interest.
I also predict the automation of all manners of finance workers and such, such as stock market brokers, insurance managers, and others that are a bit creative, but are performing low-intelligence routine tasks of high-frequency repetition. The automation of these tiresome, arduous tasks will enhance overall efficiency, and save a lot in salaries and bonuses. This has already started, here is an article about a new AI system that replaces insurance claim workers in Japan.
These sectors may be multiplied, in terms of semi-creative low-intelligence tasks, preferably with high-repetition and high-workload where benefits from AI automation will be most visible.
Here we come, Cyberia!