Will Training for New Skills Have No Value?

Posed in CWorld: An interesting premise.  As we learn to augment people with new skills using virtual assistants,  will these first replace the most valuable and difficult skills?    I covered a number of these issues in recent months while examining how segmented tasks can augment people using assistants (VDAs).  As was suggested a long time ago, is it the skilled knowledge worker that may now be ripe for replacement?  Any task you may have wanted to train yourself for,  may already be under cognitive automation.  Consider the data science case,  where there are now attempts underway to automate many elements of analytics.