Can we solve impossible problems artificially? (AI)
What role might Artificial Intelligence (AI) have on solving today’s impossible problems?
“We are continually faced with great opportunities which are disguised as unsolvable problems.”
Margaret Mead, Anthropologist
Having published Solving Impossible Problems in 2012, the nature of unsolvable problems still ‘tingles my bell’ each on a philosophical, psychological and practical level. From working in organisations and observing the world in general, there seem to be three interlinked types of ‘impossible problem’:
- Out of our control: This is a simple one… if something is outside of our control and influence, then there is nothing we can do about it. This can apply to each of us as individuals all the way through to us as the human race. Here, we can either accept our lot (even if we don’t like it) or be inspired to develop our ‘sphere of influence’ on the world out there.
- Paradoxical: This is a trickier one… when we face tensions, polarities, conflicting positions, dilemmas, competing demands, double binds, dichotomies, no-win situations etc., we often end up going around in circles, feel split or caught up in the machinery. Again, this can happen on an individual level or to a group. It doesn’t matter what we do, we either lose or end up back at square one.
- Complex: Here we face too much data to be able to make sense of it. There are too many variables for even the cleverest person to keep track of. The butterfly flaps its wings and chaos ensues, you change one thing and many other ‘unintended consequences’ are created! One person’s brain is complex, the network of human beings is complex, the humans in a system is complex, the universe is complex! We can use mapping techniques and modelling sprinkled with a dose of unconscious intuition, but it will probably be self-learning AI that will have more of a grasp on unresolvable complex problems.
I like the idea of AI tackling weather system prediction, stock market fluctuations, human group behaviour (e.g. voting), human-system interactions (e.g. traffic flow – particularly as driverless cars enter the mix). Who knows, perhaps AI will eventually crack the ultimate… how to get Southern Trains to run on time (or at all).
Computers of the past have tended to find paradox rather flummoxing. For example, put a robot in a round room and tell it to stop in the corner. Without the ability to think outside the situation, the robot will continue going around and around. Computer programmers face the constant curse of ‘code contradiction’ where an unintended loop is set up and the program gets caught in a never-ending cycle.
If AI can rise above the nature of paradox and see things from outside it’s current perspective (known as ‘going meta’), then contradiction may no longer be problem.
For me, the biggest problem with AI is the issue of control. If AI is restricted in its influence on the world out there, then it may be able to provide us with amazing insights into how to resolve the unresolvable. However, if we provide it the third ‘solution’ of control… well that may be the stuff of Hollywood futuristic disaster movies.
I agree with Margaret Mead and would paraphrase her a little, that today’s impossible, unresolvable problems are often tomorrow’s opportunities.
To the future…
(Ref: Joe Cheal (2012) Solving Impossible Problems, GWiz Publishing)
Click here to contact firstname.lastname@example.org (and to find out more about Imaginarium in-house training).
To find out more about Imaginarium Learning & Development: www.imaginariumdev.com