To optimise 1/3: machine/human learning together (with PLM as example for BI=>AI)?

Career change from BI (Business Intelligence) to AI (Artificial Intelligence) is thought reasonably as a shortcut because both are automating complex tasks - all through constant learning while performing those tasks. In Deep Machine Learning field, UL (Unsupervised Learning) stands for a new level nowadays beyond what even harder to human learning, if Alpha GO Zero would be better used in business or MedTech to clarify the patterns and anomalies that our human minds can’t feasibly find.

In order to help a MedAI start-up recently, it was very interesting to learn its clinical trial data via doing manual labelling and spreadsheet analysis, led to construct Neural Network prototype as a try. Due to the amount of data so small , as well as other start-up limits, 10 typical cases kept out especially for "independent" evaluation were difficult to proceed and conclude as the plan before this summer. ... To BI or AI, it is same how far to reach by human learning experience as the key of adventure: on its way from a few cases with the growth to big data environment.

Before moving to BI, I was working in a big company for PLM (Product Life-cycle Management) to select 3 cases as action research of PhD dissertation from 100+ product change records:

The value of operational optimisation and later job even changed to BI was virtually leading to such a way as know-how details for implementing this mission (WHY to need above 3 business or product strategies) in mind:

Now, it is the time again for proceeding AI growth to find business use case and continue beyond all limits, as human learning simultaneously to ensure excellent data / model developed in deep machine learning. In general, it can be a goal to all of industrial sectors if the future will be AI as new electricity and big data as new oil. There are the needs sooner or later if business value of AI optimisation could be determined at first!

To optimise 1/3: machine/human learning together (with PLM as example for BI=>AI)? 

To optimise 2/3: hybrid learning, developing till delivering (in ICT or MedTech context)?

To optimise 3/3: design thinking and leveraging (for health innovations)?