Some conditions don't change. An OAP with limited mobility will always require assistance day-to-day, and it's the sort of thing robots are really bad at. OAPs are also growing in number, as life expectation increases. Another one: you will always need someone to help give birth - in fact, as hospitals are deemed too expensive and scaled back, we will likely need more nurses, since home-births stop them from sharing their time between patients. Again, this will not be touched by automation. I'm sure there are plenty of other examples.

>2. How much more can they grow ?

I think that's more a function of our social organisation than technology. The real question is basically "how can we pay these people?" -- morality and common sense tell us that a lot of these activities should not be linked to profit, but that is economically challenging unless we rethink how money is created and distributed.

> We're talking about 40% of jobs at risk due to automation in 20 years.

I find that number too high to be true. 10 or 20%, maybe; 40%, it's really hard to believe - that's 15 million people in the UK without a job, where currently we have 2.3million unemployed. It would mean basically 1 million newly unemployed people each year; I would argue society would not survive 5 years in such conditions, let alone 20.

In any case, I have no doubt we can create more office-based bullshit roles to offset the bulk of those losses (dedicated entertainment people in every office; corporate psychologists; continuous-change oversight; dedicated corporate tailors; and so on and so forth...).