Humans can be Unpredictably Unpredictable - take that AI!

Humans have an ability to tackle the unknown-unknown dilemmas. We have achieved this ability by starting in the constrained known-known space. We teach our children - music class is to focus on music and math class is for math. As the domain specific foundations get strong, we let our confidence blur the lines between domains. Andy Hilderbrand's music and math domain specific foundation allowed him to unpredictably do this:

Constraints are very powerful for foundational learning - they might be the building blocks to elegantly get from AI to I for specific domains.

Empathy and creatively solving first time complex problems is an innately human trait - but that does not mean that we restrict conversational AI to the human conversation model - the model gives us a very stiff benchmark in our journey to achieve I.

The autonomous vehicle is a tough problem to solve in itself - but once we constrain it with 'autonomous vehicle in a protected lane on a highway' - the problem is practically solved ... resulting in previously unpredictable platooning solutions for vehicle fleets.

Humans uncomfortably accept unpredictably-unpredictable humans. Unfortunately unpredictably-unpredictable AI is 'uncanny valley' discomforting .. but we will get there as we aspire for bits of intelligent bytes!

Oh yes, unpredictably-unpredictable humans can be a good thing or a not so good thing - depends on the situation and your point of view.