A.I is currently only supervised A.I and is A -> B A.I so it mainly does statistics with a higher accuracy it will take a while before we can rely on it in this fashion. This is at least my opinion based on the last A.I conference, interviews with andrew ng and some of the papers I read about it.
When I was 6 in 1990 we had flying cars in some of the science for kids books. I am still waiting for them.
Although I am a fan of the possibilities of A.I I am not a hyped, we're still at the fundamental scientific research in my opinion, but lets see ... I am often wrong.
About the job trend. I personally tend to look a statistics and try to find correlation a of factors.
as a general trend towards transformation in 1 region: bls.gov/opub/ted/2016/employment-by-industry-1910…
as an indicator of growth based on our feedback in the west: insights.stackoverflow.com/survey/2018
as a general trend withing a country: statista.com/statistics/318818/numbers-of-program…
to me it looks like natural growth where I would tend to see a logistic curve: khanacademy.org/science/biology/ecology/populatio…
since it's bound towards a more biologic factor.
currently it seams linear but i don't know where on that curve I would put us. We also can see that within a 100 years complete professions shrink and others grow. So I definitely would need more data to give an prediction about our future.