Apr 13, 2017

Defying Buridan’s Ass – Factoring “Free Will” in Decision Making and A.I.

Vijay Sampath

Thanks to Sheldon Cooper in The Big Bang Theory, and the recent US elections, I was reminded of the paradox of choice, popularly known as Buridan’s Ass (actually Aristotle’s dog)

A satire on the work of the famous 14th century French priest/philosopher/scientist, Jean Buridan (He created the concept of Impetus and his work in physics, logic and economics is fascinating), the theory is explained thus-

If a perfectly rational donkey, that is equally Hungry AND Thirsty, were equidistant from a bale of hay and a pail of water, it can’t rationally choose between hunger and thirst, and because of its rationally induced indecision, it will choose neither and die!

This theory is interpreted differently for varied decision making contexts, and moral vs rational philosophical issues such as:

Over-reliance on rational thinking that leads to decision paralysis
The role of free-will in breaking a deadlock
The view that in a moral deadlock, greater good will prevail over self-interest
A situation with perfectly equal amounts of information and identical preference will cause random, unpredictable behavior
An inverse version of Buridan’s dilemma was apparently played out in the United States presidential elections.

Media reported as if the proverbial Ass i.e. the voter, had 2 unequal choices, “Good aka Hilary” vs “Bad aka Trump”. But media oversold “good” as the “rational” choice, and the Ass was informed it had no dilemma, and should go with “good”, without hesitation.

And when the unexpected happened, the “smart” people found it shocking that 59 million Americans became “irrational” and voted “bad” to victory!

I took out two learnings from this episode, keeping in perspective our increased reliance on elitist intellectualisation, artificial intelligence and machine-led decision-making and therefore a distancing from real-life interactions.

1) The constant underestimation of free will

2) The benefits of exploring the unknown, instead of seeking safety in the known or the obvious

1. Factoring free-will in deterministic prediction models

To delve into the understanding of choices, we have to factor in free-will. This in turn, is linked to the theory of determinism. This theory postulates that all actions and outcomes, especially moral choices, are based on pre-existing causes. Determinism denies the existence of uncertainty in the universe because nothing is new, therefore suggesting that free will is not possible, as all action is tainted by prior experience.

(The famous clash of these views, was Einstein’s firm conviction on the deterministic route vs Niels Bohr’s insistence on uncertainty, leading to the evolution of quantum theory.)

But quantum theory or uncertainty begs the question- Can analytics evolve from deterministic rationality to quantum uncertainty? Did lots of so-called “deterministic” voters actually exercise free-will which was based on no prior experience- After all Trump had no prior background in Politics or governance, so his promises could carry no pre-existing taints, when compared to Clinton’s deterministic lifetime record of public service.

2. Humint vs Sigint and the artifical “Intelligence” of Intellectualism

As humans cede more decisions and actions to Artificial Intelligence, it becomes necessary to reiterate the value of actual physical interaction, independent analysis, intuition and creative thinking.

Once we start relying on the results of computational analysis as the absolute truth, cognitive thinking will fall victim to the “programmed” insights of AI.

One example of machine generated influence is the current challenge of “fake news” and the reliance on “bots” to manufacture public opinion and thereby artificially move sentiment.

True insights into people come from interacting with people physically, not with their digital avatars or bot proxies. The value of body language or voice tonality in determining a person’s behavior cannot be replaced by any amount of digital tools. As they say in the Intelligence or Spy business, “Humint” (or Human Intelligence) is always more valuable and arguably more reliable than “Sigint” (or Signal Intelligence).

When a researcher visits a poor person’s home and sees the physical conditions, it should signal new insights based on ground realities or “humint”, that would not be available in digital Q&A sessions or algorithmic calculations based on a certain IPs digital behaviour.

And then there is the consensus of the elite. Are today’s homogenized, cocooned intellectual “elite” (me included) distanced from the life of the masses? Considering that our social and cultural interaction is limited to our own type, is it possible that we elites make our hypotheses and conclusions only from our homogeneous experiences? And therefore, after some time this so called intelligence is then “artificial” as it is removed from reality?

Does the outsider from the “real world” have an advantage over us?

History tells us that every advancement or mutation in human civilization has come from the repudiation of known variables, in favour of a path hitherto unknown. What is overlooked every time we take decisions or analyze complex future outcomes, is that there is a third or fourth choice, one that doesn’t conform, but offers a solution. In the case of the Ass, it was programmed to imagine that either/or was the only rational choice and that it was totally irrational to make the third choice of consuming both options in sequence, i.e. going from Hay to Water.

Our minds are heavily programmed by life experiences, fears, influencers (like Media opinion in case of the US elections) and societal norms We cannot reach our true potential if we limit our thinking solely to this “rational programming” and fail to exercise our free will, trapped as we are in our societal cocoons

Progress lies in imagining the unknown and not being the proverbial ass. Did Trump win because he knew this?

AppLy Now