2nd Law and the End of Work As We Know It - Why AlphaGo is just the beginning.
I've always wanted to figure out what it is all about. You know, the biggest picture... The forest and the trees. As a kid this curiosity of finding the ultimate truth bothered me to no end. As an adult, I figured it out. So I realized, it is all about the so-called 'second law' of thermodynamics. Which is also intertwined with the concept called the "arrow of time". Nature abhors a vacuum, we simply live in a universe (there are probably infinite universes, I'm just talking about ours) in which entropy maximization/optimization is the highest order directive. It is the only real imperative, if you will, that drives everything else. Indeed, life itself and including evolution by natural selection, etc... are all direct and inevitable consequences of entropy maximization. It is such a simple concept and principle, and yet when truly understood, it solves and answers everything.
It took all the hydrocarbon deposits (crude oil, coal, natural gas, etc) on earth billions of years to form through a slow gradual process of photosynthesis, and yet in less than 150 years since the industrial revolution began we have already consumed and burned more than half of all these energy resources in the world. There is a label and term for this, and the global phenomenon I'm speaking of is known by many as "Peak Oil". I used to wonder why we as a species couldn't be more responsible and ration our usable to make it more sustainable, over time I realized the truth, that it had to be this way, and that it couldn't have been any other way. It was quite liberating to realize and finally see what it was all about. How the world really works, with all its corruption, politics, etc is the only real way it could ever have worked.
My hunch is, due to the 2nd law, the evolution of the neo-cortical regions of our brains in the homo sapien species (aka advanced monkeys that first evolved opposable thumbs) intelligence evolved in humans because there was so much energy potential in the ground, such energy gradients required an advanced mammalian species to come along in order to truly take advantage of it and help it speed up the inevitable journey of low entropy to high entropy. The evolution of intelligence in biological lifeforms such as us and the emergence of complex systems (math, science, written language, human society, social networks/civilization, specialization of skill, economies of scale, industrial revolution, the computer age, etc) are the direct result of the impeding march towards optimization of equalization of energy gradients.
From love to lust, religion to science, computers to automation, jobs to macroeconomics and geopolitics, Darwinistic evolution to the New World Order, from Alpha Go to Lee Sedol, etc everything can be explained by this one singular force.
In the context of a dyadic interpersonal romantic relationship, Love evolved to keep two people together for long enough to give their offspring the highest chance of survival in order to pass down their genes to the future generation; it is merely one of many strategies (infidelity being another one) that conforms to the overarching darwinistic evolutionary structure of the pressures of natural selection dictating which asymmetric sexual strategies and tactics an individual uses in this survival of the fittest. Taken in aggregate, however, it merely is nature's way of letting the best rise to the top, in this larger survival of the species-as-a-whole endgame. It also exemplifies our concept of "good" vs "bad", and goes to show that all our laws, law and order, our rules and systems and practices in society and civilization exists to serve entropy maximization. For example, criminal behavior like theft benefits the individual criminal who gets away with it, but the reason bad guys get locked up (social contract/ insurance policy) is because society as a whole suffers, and society as a whole benefits from the greater good of cooperation when everyone is subjected to certain laws/norms/etc in which is conducive to creating the sort of stability that in the long run helps best facilitate the perpetual growth of society itself (and thus by extension fastest net rate of energy consumption, resource extraction, aka entropy maximization).
In our brief history we've had the agricultural revolution that allowed cities and civilizations to form (no longer did we have to be a nomadic hunter and gatherer species), and with that the development of language, specialization of skills, etc that eventually lead to the industrial revolution in which machines took over for much of human manual labor and things like the assembly line afforded us the economies of scale to cheaply mass produce goods in which served to the furtherance of growth of mankind, both in terms of global population and standard of living. Then came along the digital/computer revolution that allowed us to communicate at the speed of light (radio, telegram (not to be confused with Instagram which is an app and not morse code), television, Internet, cellphone) and very quickly process raw data that would have taken human's orders of magnitude more time and effort to calculate mentally or by hand, thus freeing us from having to do the mental menial labor and allowing us to do the more creative or intuitive things, and to focus on the bigger picture tasks, such as managing and overseeing these computer programs and systems. We are now entering an intellectual revolution of the machines, otherwise known as AGI (artificial general intelligence) in which soon will make completely redundant the need for humans to do any sort of work or hold any sort of job whatsoever. Taken as a whole, our concept of wealth is really just our societies aggregate ability to extract, process, and consume energy and resources. Once a super-intelligence comes onto the scene, this super-intelligence will be able to achieve this far superior than any human or otherwise manned workforce or society would ever hope to accomplish. 2017 marks the year that A.I. will really takeoff, for me it will be interesting and exciting to watch, from self-driving cars, to computer natural language process, soon there will not be any task however specialized or general that computers won't be able to do better than the human counterpart. Once we reach the threshold of a general artificial intelligence that becomes capable of iterating itself better than humans could do, then that sort of omega singularity will be the ushering of the second Cambrian explosion. Life begets life, evolution begets evolution, the implications of this is far fetching, but it is simply another step on this ladder of the impeding march towards entropy maximization. While we may have colonized planet earth and had dominion over all the animals, it is our descents in the form of non-biological and non-carbon based life forms such as physical digital machines that will take over the baton, it is this super-intelligent machine/digital lifeform that will one day travel to the stars, repopulate wherever it goes, evolve on its own, and someday colonize the entire Milky Way galaxy and terraform all the stars and planets.
Every religion, if you closely examine the core essence of tenets and principles, it always comes back to being about entropy maximization. Be fruitful and multiply! Even self-sacrifice is actually really just entropy maximization in another form, since "the greater good" is actually still serving the God of entropy maximization. Martyrdom boosts the meme rate of religious indoctrination which at the expense of an individual life actually helps society produce much more net effect, hence more overall total fruitful multiplying, even though the sacrificed individual can't really pass his seed anymore. Group entropy maximization always trumps (no pun intended) individual entropy maximization. This is why God tells his children to put him first, why couples should put God first in their marriage and each other second. Its all about entropy maximization.
We are now entering what is called an Intellectual Age. The age of automation and the end of work. Seen under the light of the 2nd Law and its thermodynamic imperative of entropy maximization, it makes sense that AGI and the rise of the intellect machines were always an inevitable consequence of existence in a universe finely tuned to allow life.
Currently the US Federal minimum wage is at $7.25 per hour, while the national average of gasoline is at $2.288 per gallon. A gallon of gas contains about 131.9 million joules of energy, or the equivalent of nearly two weeks of manual human muscle power and labor, given that 131.9 mega joules is about 31,491 calories and the typical human male consumes 2,600 calories per day. The minimum wage being a metric of the lower bound of what an hour of work is worth – even if it is merely putting groceries into a plastic sack at the checkout counter – and the average price per gallon of gasoline being a standard measure of the price, cost or value of one of modern societies main energy commodities, the implications are indeed far-fetching. This means that even at the most basic levels of compensation, compared to our stone-age brethren, we now enjoy at a minimum at least a 288 times energy multiple factor equivalent. To illustrate one example, the average U.S. mpg is around 25.5, and so an hour’s worth of work at minimum wage equates to approximately 3.4 gallons of gas, which is enough to move an average motor vehicle 87 miles. Since the average car can reach max-continuous speeds of 87 mph or faster, this means an hour’s worth of work from even the lowest paid members of our society will transport and propel him or her at 87 miles per hour for one entire hour for a total of 87 miles traversed. It would take days if not weeks for a person to naturally walk 87 miles, and it would take weeks if not months for that same person to push a car the same distance of 87 miles. And to look at it from another perspective, the fastest man on earth, Usain Bolt, is capable of sustaining an instantaneous top speed of about 30 miles per hour for only a few seconds, and yet the lowliest paid workers of our society can reach speeds of nearly 90 mph, sustained perpetually.
Wealth creation is actually energy extraction, conversion and consumption. The only reason the above analogies and examples are possible is due to the fact that humans have been fortunate enough to be gifted with an abundance of stored energy from planet earth, in the forms of hydrocarbons (coal, crude, natural gas) and other sources. For billions of years, through a process of photosynthesis and thanks to the natural energy from our closest star the sun, our planet has steadily built up a massive reserve of energy, mainly in the form of hydrocarbons underneath the ground. And yet in less than 150 years since the industrial revolution began, we have globally already consumed more than half of all available hydrocarbons and nearly all of the sweet crude, and other easy-to-extract low-hanging-fruit energy sources on earth. The implications are indeed foreboding.
In high school and through much of our higher education we are all taught the concept of the “time value of money”. We are told that if we save a little and put aside a portion of our money to allow it to grow (“make your money work for you”) that by the time we retire at age 65 or 70 or older, that given a reasonably sustained interest rate and steady overall rate of return of the growth of our savings due to compound interest, etc it would have amassed into a small fortune of a few million dollars or more. Putting aside the fact that the “time value of money” doesn’t take into account for the fact that a person at age 75 cannot enjoy life in nearly as gratifying a manner as someone spending the same amount for similar goods, services and experiences at age 25, there still remains the fallacy of perpetual growth. All of the above is indeed predicated upon the assumption of perpetual and continuous global growth. In pre-agricultural times when there were no forms of monetary accounting nor any surplus, we consumed what we had and when we had it, and anything extra that we set aside would either retain value or lose value over time, naturally and even from a physics thermodynamics and entropy standpoint, it would certainly never “grow” in value. So taken in aggregate and in general, the growth of an individual’s savings is directly coupled to the overall ability of society to grow as a whole. By deferring expenditures and putting that aside and allocating that portion of their monies to go towards capital to helping society and civilization to continue to expand, populate, and grow, the individual is repaid in kind at the end of his or her life as their investment in the entire pool is conflated with the overall growth of society. By putting a piece of their money into the pie, their individual slice grows larger as the entire pie as a whole expands. (A rising tide lifts all boats). Modern economics and all macroeconomic models rely upon the assumption of perpetual growth as a built-in component. From fiat currency to fractional reserve banking systems to petrodollar hegemony and quantitative easing, etc it is simply the way we have structured our society and our financial systems by assuming that we will always grow ourselves out of debt and using the assumption of tomorrow’s growth as collateral for today’s level of expenditures. The issue is modern society and thus growth as we know it to be, are both entirely predicated upon the prerequisite ability to continue to extract, process, convert and consume energy at increasingly faster and faster rates. Indeed our entire global civilization is wholly dependent upon the sort of specialization-of-skill, just-in-time logistics, and economics-of-scale that only a massive interconnected global population sustained by very high density net energy sources (hydrocarbons) could provide. Without airplanes, trains, ships, trucks and without electricity, computers, telecommunications, and the Internet etc none of any of this would be possible today. But in order to keep the lights running, so to speak, and to sustain and maintain this massively complicated global structure we need a continuous supply of high density energy sources. Mathematically speaking, perpetual growth in a finite environment (earth) is impossible, so sooner or later the pyramid scheme will collapse if society continues down its current course trajectory.
In our ancestor’s nomadic era and during pre-agricultural times, early humans spent most of their day hunting and gathering for food. During this period not only were there no concept of accumulation of wealth, but we relied entirely upon our own human abilities for survival without any sort of physical or intellectual multiplier-effects. A week’s worth of labor was what muscle and mind could produce within a week. There were no diesel Caterpillar machines for labor nor Boeing jumbojets for transport or Intel processors for compute. Everything we did was done by hand and carried on foot and decided by brain. And thus due to natural constraints the human population always remained within Mother Nature’s natural carrying capacity for our species (500 million worldwide vs the current global 7.4 billion and growing).
There is no “going back the way things were” without a massive implosion of the global human population. The only way forward in which the system and thus the fabric of modern society and civilization can maintain itself is by going forward, not backwards.
In pre-industrial times the vast majority of the population (more than 90%) was on farms and preoccupied with growing food for survival. Nowadays, hired farmworkers make up less than 1 percent of all U.S. wage and salary workers. From building bridges, to constructing walls, to plowing fields, industrial machines have taken over the need for manual human labor especially as it pertains to man muscle power. As long as it is fed an high density energy source in the form of gas or diesel hydrocarbons, a single Caterpillar dozer or individual John Deere tractor can do the equivalent muscle power work of dozens if not hundreds of men. Since industrial times and after the internal combustion engine began coming into play in earnest, these industrial machines have entirely augmented and indeed replaced the need for manual human labor.
Ever since the industrial revolution took over, human jobs and our concept of “work” has been steadily relegated to intellectual activities and pursuits. Farms still needed someone to operate and drive the tractors. Construction crews still needed engineers to design the bridges and the highways and to provide them with the blueprints and the instructions on how and where and when to build. Airplanes still needed pilots to navigate and to fly them. And so on and so forth.
With the advent of the digital and computer ages, this progressive trend continued. Once the first pocket sized calculators came onto the scene, we started increasingly load shedding and offloading the most basic elements of thinking and mental work and thus relegated them to a class of “mental menial-labor”. Simple things like mathematical addition, subtraction, multiplication and division were no longer done in the mind using mental math nor on paper and pencil with hand, it was soon simply accepted that these tasks were better left to calculators and computers which could do them far faster and more accurately, (In 1874 it took William Shanks 15 years to calculate PI to 527 digits by hand, today on my home desktop computer using a free tool called y-cruncher by Alexander Yee, I can calculate PI to several billion digits in only a few short seconds.), leaving the human free to do other more meaningful work such as applying intuition to solve problems, and being able to see the bigger picture, making the crucial judgement decisions that computers could not yet understand, interpret or adequately apply. Accountants used calculators to crunch numbers, pilots used them to calculate takeoff speeds, construction crews used them to calculate how much lumber was needed for a project, and so and so forth.
As computers became faster and software and algorithms more developed, increasingly basic automation began to come onto the scene. The invention of the computer spreadsheet for example, popularized largely by Microsoft Excel, allowed scientists, accountants and businesses alike to play “what if”, to rapidly experiment with a variety of scenarios without having to spend exorbitant amounts of time to re-compute an entire page of formulas and values. The first and most basic forms of autopilots on airplanes consisted merely of the ability to hold a set altitude or fly a defined course heading. But slowly as airplanes started to incorporate flight management computers that could automatically calculate variables such as takeoff speeds based on known takeoff weights, airport runway lengths, temperature and weather conditions, etc these decreased the pilot workload and eventually allowed airlines to entirely do away with flight positions such as flight engineer and flight navigator, items and tasks of which now are entirely automated by the onboard computers. Eventually the job of the modern airline pilot became more akin to systems management and system administrator (almost like an IT job) than it was raw flying. Most modern jetliners are capable of fully piloting themselves using automation from moments after takeoff and all the way down to touchdown, requiring the pilot only to monitor the autopilot and associated systems for when something goes wrong with the automation or during the rare inflight emergency or unusual contingency. An intercontinental flight on a jumbojet that once required eight pilots (two full crews of four each) now routinely only need three (one crew with two present in the flight deck at any one time).
While simple industrial machines (dumb machines) like trucks, tractors, trailers and dozers etc enable a man to do the physical work of many men, thus creating a force/work physical multiplier effect and displacing many manual labor jobs, in the computer age automation allowed a single person to accomplish what once required the intellectual force of many, creating an intellectual or intelligence multiplier. It wasn’t that long ago when an IT systems administrator had to build a new server he would have to go through the long entailed process of spec-ing out the hardware requirements, often building the server from scratch by procuring and then installing the processor, the disks, the memory and then spending many hours installing the server operating system, accompanying application software and standards and configuring the system afterwards. Nowadays with computer virtualization a new virtual server can be spun up at the click of a button, and the creation of a new virtual server from a template/clone is largely automated and immediate process. With virtualization (a form of automation, or at least it affords an automation factor) a single systems engineer or systems administrator can do what once took a whole team of a couple of dozen admins. Even if automation will never entirely replace the need for IT systems admins and engineers, it has already displaced what used to be or what would have been the majority of them.
Indeed, no job or industry is immune from the effects of computer automation and what is increasingly seen as “artificial intelligence” (AI) and machine learning (deep learning, or especially general-purpose artificial intelligence, neural network trained to datasets, etc). The only domains left in which humans still currently outperform computers and automation are in the realms of the creative, the intuitive, and the biggest picture decision making processes. As more and more jobs are made obsolete our definition of ‘work’ has shifted to what makes us uniquely human; things that robots and automation and computers and algorithms and machines still cannot do.
Many prominent tech companies such as Apple, Google, Facebook, Uber, Tesla, etc are working on self-driving cars, autonomous drones/UAVs and other forms of automation and bringing these things mainstream and to the popular masses. The implications and widespread and far fetching. There are now kits that can retrofit existing 18-wheelers to make them effectively self-driving. In a few years an entire trucking industry can be turned upside down on its head. Uber for example is developing and testing its own fleet of self-driving cars in preparation for a future in which human drivers will be a thing of the past, cutting down significantly on the element of human labor: drivers. (indeed, car ownership itself may soon be phased out altogether, most of the time a car is sitting there parked at the garage or at office parking, load utilization is less than 10%; the idea of fleets of autonomous cars always driving around 24-7 in circulation in a community pool of cars controlled by computers to populate locations and routes to predict and match real-time demand etc at cheaper than individual total cost of car ownership rates will usher us into a new paradigm of local/personal transportation) Amazon for its part is also testing autonomous self-flying drones for product delivery. Not only does this threatened to cut out the middle man of distribution (FedEx, UPS, USPS, etc) but it potentially reduces entire modes of transportation, even self-driving trucks won’t be needed as much when a self-flying drone can do the same and faster!
Eventually automation will reach a point in which it can automate itself better than a human can do. The developers, coders, programmers that are working on the neutral networks and artificial intelligence systems that are responsible for putting people out of a job are not they themselves safe nor immune from the same. Once general purpose artificial intelligence and machine learning can reach a point where it can develop more advanced and refined general purpose artificial intelligence on its own, at a rate and threshold better than human counterparts, then that has a sort of domino chain reaction effect, runaway automation will push machine intelligence past the realm of humans and it will be possible for the first time in human history for machines to create a superintelligence, perhaps capable of solving the deepest mysteries of existence itself, finding that long eluded theory of everything in physics and using that as a top down model for which everything else is re-built and based upon. Perhaps that superintelligence will then understand reality better than we could ever hope to grasp and make the sort of decisions for us that we have not been able to make due to political, legal, religious or other social reasons.
I think it’s worth reviewing our status quo notion and definition of ‘work’ and the associated pride and ego that we often attach to our jobs, our work, and our so-called careers. The irrational humiliation or shame associated with unemployment or underemployment or of not working and of the outdated notions of what truly constitutes a ‘job’ or what we really mean when we say we are ‘working’ … these should all be re-examined. Fundamentally, it all boils down to survival, it’s just a paycheck but even money itself is merely a means to an end.
When people refer to “work” (or a “job” etc) no one is talking about manual labor anymore. There are no jobs for manually digging ditches or plowing fields or sowing clothes or planting seeds or harvesting crops by hand anymore. There are jobs for operating machines or overseeing and managing systems that do these tasks, however. The general consensus concedes that ‘work’ is no longer purely muscle work but largely mental and intellectual ‘work’. Driving a car requires some level of refined motor-control ability and a basic amount of intellect that is independent and separate from sheer strength or raw muscle power. Flying a plane even more so... The same is true for virtually every other real job out there right now. Make no mistake, mental work can sometimes be just as hard as physical work, albeit each in very distinctive and different ways, but both require an effort to be directed towards an objective goal, be it a mental one or physical one.
I often wonder why there is this really deep seated notion ingrained in most people that it is okay for machines, robots, automation to take over for our physical work but somehow when these same systems threaten to do things that we believe should be uniquely reserved for us humans (such as creativity, intuition, the arts, song singing, painting, philosophy, writing, coding, policy making, interpretation of laws, judicial decision, brain surgery, deal making, piloting, high level business decisions, etc etc) that we suddenly take offense and go into denial or somehow pushback.
For example, we are not that far away from living in a society in which machines are capable of doing almost all of the “work”, both physically and intellectually and in the manner more effective and efficient and far more productive overall, thus leaving a large majority of the existing human population (>99%) in a position in which not only will they not have a job but also they will not even be needed nor required at all to work anymore. (unemployment as a result of automation displacing jobs is not a bad thing as long as society as a whole recognizes and implements fair policies of reallocating the efficiencies achieved by automation and jobs displaced by automation and then turning around and giving back that wealth and reassigning it back to the people whose jobs were impacted by said automation; after all it’s not exactly like the robots/machines/AI will mind not being ‘paid’ for their ‘work’ , the net total actual result for society is exactly the same, in fact I would argue it is better, since that segment of society can now be free to pursue what they really want to do as opposed of what they HAD to do….) Once artificial machine intelligence reaches a certain threshold where it can develop, create and evolve more advanced versions and iterations of itself without human intervention and do so in a manner that is better than humans could ever do, then surely it will be able to replicate itself physically as well and as well as automatically do everything that such entails, the entire vertical and horizon chains (there will be no need for a small segment of the human population to remain behind in the so-called labor pool in order to continue to work in order to “look after repairing and maintaining the smart machines”, nor to “oversee and watch over the machines to make sure they are on track”… the whole point after all is that these advanced machines will be better than us in all aspects and in all respects and thus making this a total false fallacy and redundant nonsense in the first place).
Humans will be relieved from having to shoulder the burden of having to do any form of ‘work’ whatsoever. No one will ever have to work again and that will be a very good thing for all and for everyone involved. Once this superintelligence automation society gets rolling and becomes self-sustaining it will just manage itself forever without any need for human intervention. It will become better and better at managing itself as time goes on, in fact it will reach a singularity point whereby it will achieve or approach the limit to absolute perfection. Zero maintenance or afterthought required since the artificial intelligence itself is doing the ‘after thinking’. Since wealth is really measured by a societies aggregate ability to extract, manipulate and consume energy and resources, and because machines and AI will be far better at managing and creating such high rates (thus creating much wealthier societies) of consumption this will leave the human population with unlimited free time to actually enjoy life to do absolutely whatever it is that he or she wishes to do with his or her time/life within certain limits and within his or her basic income/resource allocation by auto-society. GDP, GDP per capita, parity purchasing power, etc … it doesn’t matter what metric is used. The raw amounts of resources underneath the ground remain the same (A.I. might even solve the FUSION problem and then we have infinite free-energy and can finally stop relying on oil etc) , (no matter who or what is putting it to use, be it man or machines) and really there is no contention that AI will be able to extract, convert and utilize these resources and energy (directed towards human consumption of course!, we are not talking Terminator or Matrix here) for human consumption at rates far higher/better and more effective and efficient rates and larger total aggregate outputs than humans could ever hope to achieve, this means in such a society by all humans actually agreeing not to ‘work’ (puts “right-to-work” under a whole new meaning doesn’t it!), everyone will actually be maximally wealthy and far better off than say if a small portion or segment or group of flawed and inefficient humans were trying to stubbornly “contribute” to society by “working” in order to maintain the illusion of being irreplaceable or “doing some real good” or some other feeling-good nonsense. (it boils down to being a simple matter of machines extracting and producing goods, resources, services at a rate far faster than any manned society could ever achieve, and then turning around and intelligently distributing it amongst the entire human population in an equitable manner to all members of society, and also optimizing the structure of this hybrid society and symbiotic relationship by setting the total overall machine vs human population ratios, (even capping human population) or rather the total machine vs total human energy/resource consumption ratio in order to achieve the net effect of maximizing total energy consumption per capita (max resources per person) for the humans as hardcoded directives for the machines)
For the first time in human history, since the dawn of mankind, we are finally close on the brink to solving the deepest mysterious of existence, reality and truth, but the irony is that it won’t be mankind that does it but rather a superintelligence spawned by man’s initial creation of intelligence machines/ artificial intelligence.
Life begets life. Evolution begets evolution. Ancient Go begets AlphaGo. Once this self-sustaining super-intelligent machine society is in place it will run itself and govern itself and advance itself into perpetuity. It will spread itself to every corner of this Milky Way galaxy (maybe even discover a way to access multiple different universes!!!) while being benevolent to its human creators. This is the grand-ultimate “autopilot”. The cavemen whom first discovered fire nearly 350,000 years ago could not have imagined that they were at least in part responsible for setting into initial motion the chain of events that would have ultimately lead to this… Little did Mother Nature know that evolution of opposable thumb in a certain species of primates will someday give rise to a superintelligence capable of unlocking the deepest secrets of the universe itself. It was always about entropy maximization, and it always will be. The more the things change the more they stay the same.
Seen in that light, my post here to ask Google to be more open, and to consider open sourcing Alpha Go, is in my opinion in perfect alignment with the thermodynamic imperative of the universe itself. viewtopic.php?f=10&t=14121