Friday, December 09, 2016

ARE YOU AN ALGORITHM?


Probably for as long as there have been humans, they have worried about whether they were unique.  Are there aliens just like us?  Are those people who seem to be so different actually just like us?  Are men like women?  

We’ve been particularly anxious to distinguish ourselves from animals.  This essay by Antone Martinho addresses that in a careful and amusing way.

“The Edge” and Aeon often seem to be in oblique conversation, considering the same trendy subjects from different points of view.  This “Edge” essay by Brian Christian linked below discusses humans in comparison to machines, so that with the previous essay it forms a kind of bracketed pair for a person wanting to reflect.

Each of these men chose a focus on thinking that is suitable for their own version of the problem:  ducklings for the man distinguishing the animal from the human (with a sidebar of little parrots) and computer-handled data (algorithms) for the one learning about machine “intelligence.”

Neither or both is considering human introspection which is what most thinkers are really using for both defining the concept (consciousness IS introspection) and for dealing with it (reflecting on the subject).  Both use video to pin down behavior so the same thing is observed. 

A third consideration is represented by an Aeon essay by Raoul Martinez, one which is practical, even crucial, because it asks what will give us the information of whether a criminal is likely to reoffend.  The conclusion is that what works is a mix of human judgements and data — an algorithm combining history and dimensions of both “type” and individual.
utm_source=Aeon+Newsletter&utm_campaign=87302bfc1b-EMAIL_CAMPAIGN_2016_12_08&utm_medium=email&utm_term=0_411a82e59d-87302bfc1b-68600437

The claim has even more crucial relevance when we think about something like the recent presidential election.  Should someone have designed an algorithm for presidential qualifications?  Would it best include information about success in life, about personal success, about propriety in dealing with others, about age and experience?  What weight should military service have?  Should medical fitness be included, esp. the state of the brain and heart?  

And then the scientist nods to the effect of certain rhetorics and why they are persuasive even when they are unsupported or ridiculous.  Why is an emotionally appealing falsehood given much more weight than a dry truth?  He concludes that “human consciousness” is full of bugs.  This judgement is reached by philosophically considering computer science data.  Like the other scientist, Antone Martinho, who philosophically considers the data he has accumulated from getting imprinted ducklings to follow a dummy mother around and around a labyrinth.  Should he throw out aberrant ducklings, like the one who got fascinated with the rotating arm mechanism of the dummy or the one momentarily confused by a noise?

There is a profound relationship between the biggest philosophical questions and the rigorous mathematical thought as done by machines, particularly when it comes to ethical questions, as framed up by duckling-experiment strategies.  The problem with making God-like decisions is that no one ever knows enough to really have a grasp of the consequences.  What if killing Hitler caused him never to have a son who, motivated by horror at his father's actions, would make a world-changing decision for the world as a whole?  Maybe even finally understand how to erase the very racism that caused genocide.

In the current news is the decision of a dictator to erase drug problems being caused by the whole network of distribution by simply shooting all the drug sellers.  He can do it and he is.  People express that opinion, but rarely has anyone had the power to actually DO it.  Now we will be in a position to see whether it makes a difference.  The cynical will say the vacuum created by the deaths will only be an opportunity for people who previously were kept out.  The sociologist may say that it would do a lot more good to address poverty, sanitation, hunger, education, and all the other algorithmic causes of commerce outside the law.

What if shooting all the sellers is so effective that it’s decided to shoot all the buyers as well?  Or to decide that paid sex is just as evil as drugs, justifying on-the-spot murder.  Why waste time and energy on measured justice?  Why even trouble the citizens with the decision?  We are told that in some South American cities the cops simply shoot street urchins on sight, though they are hardly more criminal than rabbits.  Of course, tyrants have always assassinated their rivals and critics.

Extremely harsh punishments are supposed to discourage behavior outside the prescribed limits, and yet most studies of data suggest that the most humane kinds of response work the best to redeem offenders.  They just don’t feed the hunger for revenge that can haunt society, causing the mass of citizens to become criminal, as in Nazi Germany or among Islamic extremists.  Or American lynch mobs.

Two factors seem to predict this turn of events:  a concentration of wealth and power in the hands of a few, and secrecy/“opacity”, which means it’s possible to deny plain and obvious facts.  Our justice system is based on our ability to explain ourselves when it comes to motives and actions.  Lawyers are not detectives:  they take the evidence they have and explain it in the most flattering way.  (Our current cultural problem is that people invent evidence, the “false news” dilemma raised by social media which claims deaths and phenomena that are fictional but disruptive.)

What is rational?  Brian Christian suggests provocatively that It is NOT accuracy or reliability that are rational.  Those are a luxury.  Rather it is eloquence that makes the decision happen.

There is an increasing body of evidence about how humans make decisions.  “Game theory” is influential for those who even know it exists, but probably most people think of games as “sim”-reality, simulated by all-seeing designers.  The minds of game players are shaped by this conviction.  

Christian brings up “restless bandit” theory, in which a compulsive casino gambler goes up and down a row of “one-armed bandits,” each programmed to make payouts at a different rate, which seems highly relevant to the recent election.  But maybe it just boils down to the emotional theory that it can’t get worse, so throw the bastards out.  Change to a new machine in hopes of a better outcome, never thinking of leaving the casino as an unreal context.  

Martinho testifies that there’s always the rare duckling who is able to jump the wall of the labyrinth.  We are not all just like each other.  We are looking for the exceptionally eloquent.

No comments: