Artificial Philosophy

In machine learning, there is a training period that rewards or punishes a machine for its choices when given a set of training examples.   In the process of learning, the machine adjusts its internal state to better recognize the desired outcomes while reducing the occurrences of disagreeable outcomes.    In the case of neural networks (such as in deep learning) the internal state of the machine is a large number of metrics that it uses to combine inputs from the previous layer in processing.   The totality of the layers of nodes and the weights between the nodes in each layer capture whatever it is that is being learned.

This process of learning by example is one of the methods for achieving artificial intelligence, but to me it seems better described as artificial wisdom.   The wisdom is making choices based on lived experiences.   We can increase the wisdom of such a system by adding more layers or more nodes so as to benefit from ever longer training sessions.

Once properly  prepared, the machine can begin to take over tasks previously expected from trained animals or even humans who act as assistants to a human supervisor or manager.

As time goes on, it becomes more cost effective to add more learning depth to the machine and thus even longer and wider training examples.    The result is something that can replace even more senior human positions where the qualifications or more likely to be described as wisdom instead of intelligence.

At least for neural network type machine learning, what is being learned is wisdom.  I guess the word intelligence applies to the capacity of being able to be taught, but I think the word intelligence includes the capacity of innovating with the result of having new things to be taught.

My presumption is that intelligence and wisdom is not unique to humans, but instead is pervasive throughout all living beings.   I do not presume these concepts do not require a brain, let alone a human one.   A human brain is specialized to concentrate both wisdom and intelligence into a single individual that can quickly access this faculty for decision making with beneficial consequences for that individual.

As humans become more inter-connected into larger populations with quick communication between people, there emerges a more common manifestation of intelligence of hive minds, and even of multi-generational species-wide intelligence similar to the concepts of animal spirits.   We collectively make decisions that have wide ranging consequences for the entire population even when the decision can be detrimental to the one making it.

As time goes on, the overall behavior of the human population on earth begins to resemble the intelligence and wisdom of lower animals like insects.   In particular, the intelligence and wisdom of the individual matters less as technologies improve and the human population experiences both growth and increasing interconnected communication of ideas and consequences.

During the transition I’ve personally experienced over the past half century, I sense something being lost.   There is something we valued more in a lost but better time.  Yet if asked what I would like to recover from the past that is no longer available, the best I can come up with are ecological ones: examples where nature was more untouched by man while in some sense humans were not as well off.   Certainly, there is a strong case to be made for ecology, and in particular, to make sacrifices in the human economy for the benefit of nature.    However, this is not a sufficiently satisfying answer to what I feel is being lost.

Once I renamed machine intelligence as machine wisdom, I saw the connection.   We individually gain wisdom through learning, especially through experiences.   This has been part of human nature for all time, and I presume it is an aspect of all animal life even as the experience may be shared over populations or generations instead of individuals.

For humans, this individual wisdom allowed us to progress up through the early iron-age with large scale civilizations that arose in part by recognizing particular leaders deemed to have superior wisdom, in the sense of being best able to make decisions for the population.

One of the contemporary controversial questions is why is western civilization so much more successful than other civilizations, past and present.   There are many conflicting explanations that focus on different characteristics of what makes up western civilization.   There is also the difficulty of a long and inconsistent history of this civilization.    I described this earlier in the example that I’m sure my ancestors would not appreciate having me as a descendant.

Western civilization even from 200 years ago is very different than it is today in too many ways to count here.   Yet, I propose a particular definition of western civilization as an invention of the Greek philosopher Plato.   That invention was the concept of philosophy.   Philosophy is the love of wisdom but where that wisdom is expressed publicly with the intent to persuade but in a way that subjects that same wisdom to criticism.   It may be unfair to credit him with the invention of philosophy, but his surviving works presented us with a definition of philosophy that include challenging the naturally arising wisdom each individual builds over a lifetime of experiences.

Over the 24 centuries since Plato’s contribution, western civilization has experienced many collapses with devastating consequences.  By western civilization, I’m referring to the history of mostly the European continent influenced by Roman empire.    Based on this history, this group of people does not appear to have any inherent advantage.

Despite that history, I see Plato’s contribution as being very influential in what happened afterwards, and that influence is generally beneficial.   Again, I see his primary contribution in defining philosophy as a love of exposing ones internal wisdom to others.  He had a particular set of wisdom that he did his best to argue, but his more important contribution was to celebrate the dialog of different people discussing their privately acquired wisdom.   The dialog is what mattered.

Internal wisdom is a result of learning sometimes through direct instruction by a teacher, but often acquired through personal experiences such as through apprenticeship.   This is analogous to machine-learning that has led to recent accomplishments.   Machines are building an internal wisdom that is at least an analog to human internal wisdom.

Given the pace of advancement of machine learning, it is easy to envision a future where everything is run by machines rather than humans.   I believe machine intelligence can take over all human occupations.  We’ve discovered the technology for building internal wisdom that we once could only expect from humans or trainable animals.   It is easy to imagine a future where the only jobs available to humans would be the hazardous work such as those involving first responders to unexpected disasters.

Machine intelligence can replace most jobs, including the professions requiring years of specialized training rewarded by secure and comfortable jobs.  Such technology can eventually run entire governments where the chief executive is a machine.

One of the themes of this blog is to explore the consequences of a type of automated government, one that is run by data.   Such a government would be very different from what we know today, but one with both benefits and disadvantages, and I think we’ll learn to accept the disadvantages of a potentially harsher government.

In my earlier posts, I extrapolated to what a government by data would be like.   I kept coming to the conclusion that it will begin to resemble ancient civilizations.   In such civilizations, the populations accepted the supremacy of the ruler and acquiesced to his rule in order to receive allowances for subsidence.   The future automated government will be similar, the people will accept the rule of a machine and rely entirely on that machine to give them their allowance, something already discussed as a guaranteed income in exchange for obedience to the law.    Even though the government will be based on algorithms and machine-learned wisdom, I predict that government will resemble the ancient civilizations in terms of how the population is treated.   Basically, the government may be largely libertarian as long as people stick close to their assigned roles but will be ruthlessly authoritarian against any attempts to stray from those roles or territories.

The future of an automated government by the current generation of machine learning (wisdom learned from training) is going to be a reversion to an older model, but one that may work better now due to the technology being supreme at handling human disorder or rebellion.

This future government will lack the qualities of a western civilization because its rulers rely entirely on their own privately acquired wisdom, even if that wisdom is of very high quality.

Machine learning technology has demonstrated that we now know how to automate wisdom acquisition in machines.   We can replace humans with machines where the job requires only this type of wisdom.    Most jobs will fall into that category.

As far as I know, machine learning is not able to do philosophy, however.   In particular, a machine is not able to express its internal wisdom in such a way to engage in dialog with another owner of internally acquired wisdom so that the two forms of wisdom can compete with each other.

I’m thinking particular of those various weights between nodes of different layers in neural networks.   Different instances of the same network of nodes and layers will end up with very different patterns of weights when given different training and yet each can be competitive in terms of completing the same task.   Many different systems of wisdom can perform the same task to a similar level of effectiveness.

Until machines can express their internal wisdom to have a dialog with others expressing their internal wisdom, the resulting system of behavior would not be “western” by my definition.   Certainly, we can mimic every aspect of modern life with automation, and the resulting behavior can be perpetual as long as there is no catastrophic change to the facts on the ground.

As I described in an earlier post, a catastrophe can arise from something as simple as a change in demographics where one population out-breeds and out-lives a neighboring population.    The automated government based on internally-learned wisdom can definitely respond to such changes, but it will do so in a way to attempt to maintain order.   It will not be able to innovate a new order that departs from past wisdom.

In my thinking, the primary advantage of western civilization is its celebration of the concept of philosophy set down by Plato.   This concept is that we expose our internally acquired wisdom to our peers who have developed different wisdom while being equally able to participate in society.   The love of wisdom of philosophy refers to a form of love that puts that wisdom on display for others to absorb, and this inherently presents a conflict between different models of wisdom.   The resulting debate exposes the wisdom that may be better at addressing the current circumstances.   As a result, it is this idea about philosophy, not wisdom itself, that drives progress and the success of western civilization.

While western civilization generally celebrates its philosophers, there are many times of stability where certain models of wisdom suppress the expression of new wisdom acquired by individuals.   This stagnation sets up the civilization for at least partial collapse during which time the previously suppressed wisdom of individuals are expressed.  During these times, the many competing models of previously suppressed wisdom hastens the collapse, but often a new consensus emerges that renovates the civilization to best take advantage of the current circumstances.

While much more work is needed for machine learn to match human capacity in the most intellectually challenging jobs, it is not enough to match human potential for learning.   Matching effectiveness of western civilization requires automated philosophy, the machine’s ability to express its internally learned wisdom to communicate to others (man or machine) with an invitation to have a dialog to persuade or be persuaded.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s