The Design Flaw in AI?

I've been captivated by https://www.deepmind.com/ of late.  Well not so much captivated as motivated to think about it more deeply.

Since I was a middle-school student I was a student of science-fiction--  Became an avid Asimov fan the first time I read Foundation and the aspect of the singularity occurring in so much of the classic SF literature certainly captures my imagination.

I'm a teacher now.  Sort of, or rather - of sorts.  I noticed something this morning that I felt compelled to share because I want to know more about it, talk to more people about it, and I see something I've never seen before:  There is a way past the singularity.

If you don't know what I'm talking about when it comes to AI and the singularity you can go here for some context: https://en.wikipedia.org/wiki/Technological_singularity

The Telus School for Innovation by Design started up this year as a collaboration between Calgary's Spark Science Museum: http://www.sparkscience.ca/ and The Innographer https://theinnographer.com/ and I was welcomed as a member of the inaugural cohort.  As a teacher, this has helped me understand what it is about the Design Process that works (On Design Thinking: http://www.creativityatwork.com/design-thinking-strategy-for-innovation/ )

The fundamental take away, for me as a teacher, has been the incorporation of the double-loop learning model pitched as the energy behind the machine of Design Thinking.  It is in this that I see a path around the singularity:  Empathy.

Here's the quick and dirty of it:

Defining the Problem
HL approaches me looking for a way to reward the students who won her robotics competition.
I listen and connect to what it is she is looking to do.  Ultimately it comes down to using the new school 3D printer (An Ultimaker 2 Extended!) to do something for the students we've never been able to do before.

Ideation and Pitching.
I take the idea away and work up three potential solutions to present to her (informally)

Iteration
She chooses elements of two of those ideas that work best.  Taking advantage of our growing local Maker economy using coins for PLA we will be supplement the winner's Maker account by giving them a unique coin that can be cashed-in for an extra 150cm of plastic.

Prototyping
I work up the prototypes and have them ready for when she needs to gift them to the students.

Critical Feedback.
During this phase I have to learn the programs, rely on help from local experts and mentors, and design and redesign as I look for feedback.

Design Thinking:  Better results than I would have started with because it involved the work of many for the benefit of many.  More than that, I became more skilled AND more knowledgeable because of my work with technology.

According to the Innographer - this should count as Double Looped Learning: https://en.wikipedia.org/wiki/Double_loop_learning otherwise known as the engine that drives the success of Design Thinking.


Learning is not a product based economy.  Our schools often result in a product based economy because it's efficient, mobile, and universal.  Unfortunately it's why so many have argued that the Industrial Model of education is broken; learning is a process.

In my very limited understanding of the work of AI scientists I see us still living in the industrial thinking model.  At it's best - what can we make faster, stronger, and more efficiently, at it's most basic - how do we program the machine to win?  A very industrial view.

Back to the Point
What strikes me is the story of Fan Hui (really great Wired article:  http://www.wired.com/2016/03/sadness-beauty-watching-googles-ai-play-go/).

If the purpose of AI is to create a machine that learns in order to create a product then soon the product may come at the expense of the person.  This is the technological singularity that we fear.

If we employ what happened for Fan Hui - i.e., the machine learns in parallel with the improvement of the human, then we have a) empathetic learning and b) no risk of learning in order to win.

Perhaps it's a bit socialistic.  Perhaps it's Utopian.  What if our machines learned to be our companions, who pitched prototypes of their ideas to us and then relied on our intuition and interaction to co-develop the processes that would lead to the product?  That's the discussion I want to have.  I think it will improve schools, and I think it will accelerate AI.  That takes research and smarter people than I... but I suppose it warrants thinking about.

Afterall, all of us are smarter than one of us... even if the one of us is the world's smartest computer.