The Advent of AI And the Changing Nature of Work
Should humanity be concerned about singularity? Ray Kurzweil does not believe so. In fact, he is unconcerned. He claims that what science fiction depicts as a singularity — when AI enslaves humanity — is fiction. "That's not possible." According to him, singularity is an opportunity for humans to improve.
"I have set the date 2045 for the 'Singularity,' which is when we will multiply our effective intelligence a billion fold by merging with the intelligence we have created." —- Ray Kurzweil.
Ray Kurzweil, inventor, futurist, and CEO of KurzweilAI, foresee a future synthetic neocortex engineered with nanobots to amplify brainpower:
Twenty years from now, we'll have nanobots—another exponential trend is the shrinking of technology—that goes into our brain through the capillaries and basically connects our synthetic neocortex and the cloud, providing an extension of our neocortex. Today, you have a computer on your phone, but if you need 10,000 computers for a few seconds to do a complex search, you can access that for a second or two in the cloud. In the 2030s, you'll be able to connect to that directly from your brain. I'm walking along, there's Chris Anderson, he's coming my way; I'd better think of something clever to say. I've got three seconds—my 300 million modules in my neocortex won't cut it—I need a billion more. I'll be able to access that in the cloud. Our thinking then will be a hybrid of biological and non-biological thinking.
Moore's Law Becomes a Misdemeanor
Here I interrupt the majestic trajectory of Kurzweil's thought with an observation about diminishing returns as they involve the escalation of computational power. Moore's Law, named for Intel co-founder, Gordon Moore, states that the number of transistors in an affordable dense integrated circuit will double approximately every two years.
Note that Moore's Law, like the concept of EROEI, incorporates affordability as an essential dimension for understanding. The exponential growth of computational power, up about a million-fold, since Moore formulated his Law nearly half a century ago, would not have been so exciting if the cost of acquiring this computational ability had multiplied by even one thousand.
That would have prohibited developments like the iPhone, the iPad, and the wide dispersal of computational power. My son Arthur would be spending more time reading books and less time playing Minecraft. Recently, the doubling period has stretched out to about 2.5 years. In other words, this implies that the progress in the world over the next decade attributable to extra computing power will be just half of what it otherwise would have been.
Among many implications, Ray Kurzweil's projections in The Age of Spiritual Machines (p. 3) that the operation of Moore's Law will result in "computers achieving the memory capacity and computing speed of the human brain around the year 2020" now have been postponed, along with the singularity—the moment when the rising intelligence of computers leads to a merger of man and machine—Kurzweil had that sketched for 2045. But if computational compounding is halved and then halved again in the next three decades, we may have to think more for ourselves.
Kurzweil is a man of formidable intelligence with a record of success in technology development—he foresees hard AI within twenty years. Even if the date proves to be postponed, as Peter Thiel suggests, this is equivalent to a credible forecast that aliens from space will land on earth within the foreseeable future. What would you think if you knew that flotillas of flying saucers soon hover overhead, looking for a good place to park?
Would you be worried that space aliens could do your job more cost-effectively than you? Would you be concerned that they might eat your children? Or would you wish to join the welcome party and host a celebration in their honor?
The implications of, and reactions to, the advent of hard AI covers a similar range of possibilities—almost the whole inventory of science fiction plots from the past half-century could come to life. Rather than trying to rehearse them here, suffice it to say that the implications would be far-reaching and disruptive.
Likewise, the possibility of artificial life, whether it emerged from the evolution of molecular-level assemblers or through the programming into machines—the downloading of human consciousness—as imagined by Kurzweil, could have astonishingly far-reaching implications. These would include the probable emergence of artificial organisms that could do our physical work and collaborate in problem-solving.
If these artificial organisms were considered robots, they would displace the need for much human work. If they achieved quasi-human status, they might become varieties of enslaved people. At the very least, this would suggest the emergence of a complex caste system, in which a range of creatures reminiscent of the bar scene in Star Wars jostled for legal status. The puzzles and quandaries this would entail seem sure to perplex the future.
The ambition among researchers to decipher the most intimate secrets of biology includes a desire to decelerate significantly or even transcend death. The aforementioned Kurzweil not only forecasts that hard AI will be with us within twenty years, but he boldly believes that the exponential progress of information technology will make heretofore unimaginable miracles possible. As reported by Caroline Daniel in an April 2015 Financial Times article, Kurzweil declares that humans will overcome almost all diseases and aging over the next twenty or twenty-five years.
This is a notion about which I am optimistic. The revolution in molecular medicine is based not only on the expansion of computational power, as emphasized by Kurzweil but also on the 400,000-fold improvement in the magnifying power of microscopes in the centuries since Robert Hooke first discovered the cell. Hooke, under commission to King Charles II, discovered the cell with a compound microscope with fifty times the human eye's resolution. Today, the new STEHM Microscope allows researchers to see with a resolution of twenty million times human sight. Science now has the ability to see material as small as a single atom—a million times smaller than a human hair. The deepest secrets of life, literally hidden by invisibility since the dawn of time, are now open for inspection.
My old Oxford mate, Matt Freeman, grew up as one of the world's experts in rhomboid proteins. More impressive, he is the head of the Sir William Dunn School of Pathology at the University of Oxford. (That is where penicillin was developed s a therapeutic). Matt believes that the incredible increase in telescopic magnification means that we are at the threshold of great flourishing in the effectiveness of regenerative medicine.
Some animals, such as lobsters, whose cells are amply supplied with telomerase—the enzyme that preserves the length of telomeres across cell divisions—do not die of old age. A lobster could live hundreds of years old if it could elude the lobster pot, large fish, octopi, and cannibalistic encounters with other lobsters.
Unfortunately, in humans, the telomerase enzyme is usually found in profusion only in the telomeres of chromosomes in germ cells and some cancer cells. Although there are complications, the biology of telomere extension suggests that it could be a potent mechanism for extending the human health span.
Google has created its own venture to combat aging, Calico, to which Kurzweil is an advisor. Whether it is Calico, or something entirely different, it would seem to be only a matter of time until the exponential progress of information technology leads to effective therapies to supersede our inherited biology and counter aging.
Biotechnological breakthroughs that merely permitted greater numbers to reach a decrepit old age would have very different megapolitical implications than innovations that counteract the aging process itself, reprogramming the body to reverse senescence. For the sake of this thought experiment, imagine that the highly disruptive innovation of "transcending death" only has the effect of doubling the human health span. Think of people enjoying the equivalent of additional three-quarters of a century of vigorous living in what would be a health state equivalent to being in their forties but protracted for decades.
Dr. Al Sears shrewdly observes that a significant extension of the health span would open the way for expanding human capital. You would invest more in developing yourself if you could anticipate many more useful decades of life. Today, people tend to "grow up" in their early twenties, after which they more or less attempt to make the best of the person they have become at that point.
Learn Chinese? Why not? Become an elite athlete? Build a physique like Dwayne "The Rock" Johnson? With enhanced insights into health and fitness, you could become a fair semblance of a superhero with an extra few decades of vigorous life to work on it. Think of the possibilities. Even if you were not an athlete at twenty-one, you could become one at eighty-one, when the scientific basis for reprogramming your body to grow younger is better understood.
Creative Destruction Through Survival
You cannot consider the implications of a potentially disruptive innovation like the conquest of aging without recalling that the current corporatist system feasts on the diseases of aging. The pharmaceutical industry, along with the whole sick care establishment, has trillions at stake in the continuation of the current system.
You know what they say; "Life follows art." With that in mind, you might find a template for grasping the impact of a biotechnological breakthrough to overcome aging by watching The Fiendish Plot of Fu Manchu, the last film completed by the late comic Peter Sellers. In this flick, the wicked Fu Manchu is 168 years old, courtesy of his elixir vitae, a secret anti-aging potion known only to Fu. You could take Fu Manchu to a fictionalized version of Li Ching-Yun, the Chinese herbalist who died May 4, 1933, supposedly at the age of 256.
If there is a major life extension breakthrough, I would not be surprised if it turned out to be of the Fu Manchu variety: a secret elixir available only to drug dealers and major players in the white slave trade (or maybe some more respectable billionaires), rather than a common tonic administered through systems of socialized medicine. The last thing Social Security could afford would be general progress in life extension. Another century of average life would bury the whole gag. Therefore, I would not be surprised if governments sought to suppress therapies that effectively counter senescence.
Needless to say, pay-as-you-go old-age benefits like Social Security could hardly survive the transcendence of death. The prospect of extended lifespans through biotechnological innovation is only one of many factors hinting that saving is destined to make a comeback as the principal expedient for addressing life's contingencies.
In all probability, any appreciable progress toward transcending death would rapidly retire the idea of retirement. (That is only one sense in which the transcendence of death would curtail incentives to procreate.) Removal of death as a universal social laxative would clog the advancement channels for young people. It would simultaneously open new categories of problems heretofore unknown in society.
Another logical consequence of the dramatic extension of human health span would be an increase in the physical risk-aversion of the rich. Statistics on "life expectancy" are easier to come by than for "life span." Even so, from what we know, it appears that a twenty-year-old man on the eve of the Civil War in the mid-nineteenth century could have expected to live another 40.1 years. Now imagine the prospect of trebling that life span at a higher level of health. The desired result would be a greater reluctance to undertake the hazards of battle, as persons with revamped biological processes would have four times more to lose than soldiers with far shorter life spans in the past.
Extended Life Span, Time Preference, and Capital Formation
A framework for understanding time horizons in the economy is the concept of time preference. Time preference refers to the relative valuation placed on a good at an earlier date compared to the valuation of that good in the future. An individual with a high time preference is one that has a strong preference for immediate gratification and a lesser inclination to invest and defer consumption to the future.
Think of Aesop's tale of "The Grasshopper and the Ant." The grasshopper had a high time preference. The ant evidenced low-time preference through its inclination to save.
Investopedia gives this capsule background on the concept of time preference: "This theory was initially constructed in 1871 by Carl Menger, an Austrian economist. This theory also stipulates that the consumer's rate of time preference, and therefore the interest required, will probably rise as the consumer's savings increase. This means that the consumer is likely to restrict their savings to a level at which the rate of time preference equals the rate of interest paid on savings."
A pronounced effect of extending the human life span would be to lower time preference, stimulating more future-regarding behavior (more savings and investment). This implies more capital formation, higher productivity, and higher living standards.
The Austrian economists, who are connoisseurs of time preference analysis, argue that higher levels of investment lead to what Murray Rothbard described as the "free ride" from the actions of others. They associate civilization with falling time preferences and decivilization with high time preference for immediate gratification over future consumption.
The basics of life dictate that you must consume as long as you live. To this extent, at least, everyone has high time preferences at mealtime. As Robert F. Mulligan points out in his paper "Property Rights and Time Preference," however, the general level of time preference in an economy has far-reaching implications. Mulligan says that only once people began valuing and recognizing property rights could they utilize more productive roundabout production methods. He also says that artificially imposing higher time preference "creates incentives for collectively undesirable behavior." In this sense, if humans, like lobsters, did not die of old age, this would increase the payoff for avoiding conflict and lengthening the production structure.
Production and Plunder
Professor Hans-Hermann Hoppe, the distinguished Austrian economist, is more emphatic in describing government efforts to raise time preference as decivilizing forces artificially. He was not referring explicitly to the Fed's QE and zero interest rate policies (ZIRP). Still, the logic he spelled out clearly puts those policies in perspective as "decivilizing."
In "Time Preference, Government and the Process of De-Civilization: From Monarchy to Democracy," Hoppe points out that government political takings, which I refer to as "plunder," affect time preference systematically different and more profound way than crime. He explains that such plunder interferes with private property rights, reducing people's supply of present goods, thus raising their effective time preference rate. People will then associate a permanently higher risk with future production and adjust their expectations downward regarding the rate of return on future investments. This means that "their expected rate of return on productive, future-oriented action is reduced all around, and accordingly all actual and potential victims become more present-oriented."
Furthermore, suppose these government property rights violations grow extensive enough. In that case, humanity will cease building an ever-growing stock of capital and durable consumer goods. They will become increasingly shortsighted, and their distant goals may not only cease but be reversed toward decivilization. Hoppe states, "formerly provident providers will be turned into drunks and daydreamers, adults into children, civilized men into barbarians and producers into criminals."
Time Preference and Comparative Advantage
The idea that shifts in time preference civilize or decivilize societies is closely correlated with the process by which comparative advantage enshrines or destabilizes property rights, thus shifting time preference predictably.
Robert Mulligan explains that individuals with comparative disadvantage in production are motivated to use violence against those with a comparative advantage in creating wealth. He points out that "in an environment where property rights are insecure…individuals will have short time horizons and high time preference."
Suppose even a few jurisdictions can efficiently protect their inhabitants from violence; in that case, the result could be an unprecedented flourishing of civilization. Thus the extension of human life span—promised by the incorporation of digital information into the production process—implies a civilizing counterbalance to its impact in increasing the "comparative advantage in violence" of unskilled persons, as well as the effects of systemic chaos in raising time preference.
Obviously, there are other pregnant issues associated with the taming of death. There is the question of how access to the biotechnologies involved would be priced and shared. If access were less than universal, there would be security issues larger than those at present about how those rich enough to afford these life-augmenting therapies could physically protect themselves from others who could not afford them. People would no longer be dying so frequently from what we now know as "natural causes." But they would presumably still be vulnerable to accident, infection, murder, and another external injury. It is reasonable to foresee that, unlike the case during the industrial epoch—when low-skilled persons employed the leverage of violence to make hostages of industrial facilities—in the Information Age, it is more likely that entrepreneurs will be targeted as potential hostages.
Thus the advent of the Information Age poses the prospect of another epidemiological transition in human history. The primary cause of death in hunting and gathering societies was external injuries. That may be the case again. The advent of agriculture brought infectious diseases. The Industrial Age saw cardiovascular disease emerge as the primary killer. If the chronic diseases of aging and replicative senescence were conquered, the remaining candidates as the major causes of death would be external injuries, infectious disease, and perhaps the cumulative effect of radiation exposures. Such a sweeping change would leave you a lot to think about.
The Changing Nature of Work
Industrial production standardized products, as well as work. Assembly lines were designed so product quality would be the same whether they were manned by geniuses or unskilled persons of modest ability. Because most people had similar marginal productivity, their incomes were similar. Consequently, the factory system gave rise to narrow income dispersion or historically low levels of income inequality.
By contrast, the characteristics of information technology are very different. They encourage income dispersion. For one thing, unlike mass production products, such as automobiles or refrigerators, the marginal costs of many digital goods are vanishingly small. This means that, unlike industrial products, the capacity constraints on digital goods are immaterial. As Erik Brynjolfsson and Andrew McAfee put it in The Second Machine Age, "A single producer with a website can, in principle, fill the demand from millions or even billions of customers."
The fact that a great amount of value can be created with few employees and without production being rooted in any particular locale has far-reaching megapolitical implications.
For one thing, it emphatically gives the lie to the Labor Theory of Value. We can now easily see that value is created by the entrepreneurial imagination and not necessarily by sweat and toil. The hordes of industrial workers with whom Marx identified are literally superfluous to many high-value operations.
The fact that many information products can be delivered digitally also reduces the leverage of violence in redistributing their rewards. For example, disgruntled truck drivers or dockworkers cannot blockade the loading docks to the World Wide Web. Short of shutting down the Internet altogether, bullying can play little part in determining the compensation derived from selling digital products.
Achieving a relative advantage in digital commerce will lead to domination. Even among computer programmers, however, a probable increase in income inequality in the Information Age can be illustrated. Think of comparing two programmers: one may create an algorithm that is essential in robotics and worth millions, while another, who has worked with the same equipment, hasn't come up with anything nearly as valuable. Over the past fifteen years, the income gap between top programmers and the great majority of run-of-the-mill programmers has widened further than your typical union organizers could bring themselves to admit. In The Second Machine Age, authors Erik Brynjolfsson and Andrew McAfee report amazing statistics compiled by the New York Times showing that three-quarters of software developers in the "app economy" made less than thirty thousand dollars. In comparison, only 4 percent of them made over a million dollars.
The digital economy works against economic equality. Not only does it mean that the top performers can earn incomes many times higher than most people's, but it also undermines the egalitarian conceit in other ways. For example, today, due to individualization in information technology, it is evident that programmers who write codes for products rely on their own skills—others cannot claim credit or take responsibility for success as easily as they may have been able to in the past. This also suggests why labor unions play little or no role in the leading companies in the digital economy, apart from a few "old economy" shuttle bus drivers who ferry high-tech Apple, Microsoft, and Facebook workers around San Francisco and its environs.
Another implication of the growing integration of digital information into the production process is that—short of hard AI or artificial life—a great many old areas of employment will be subject to dramatic change. An outline of what to expect was provided in a 2013 Oxford Martin School report, "The Future of Employment: How Susceptible Are Jobs to Computerization?" by Karl Frey and Michael Osborne of Oxford University. They estimate that 47 percent of US jobs are vulnerable to being substituted by computer capital within the next twenty years.
Of course, these projections are subject to the discount that they imply something remote from the facts—productivity growth should have been skyrocketing in recent years. But it has not. As you know, productivity is measured in human labor output per hour. If robots had taken over any considerable fraction of production, this would have caused a surge in productivity that is not in evidence.
That is not to say that robots and AI will not take a bigger role in automating jobs. But the projection is not as convincing as it could be if productivity were already surging. And this, indeed, maybe a message confirming that the government has been telling prodigious lies in reporting employment growth in recent years. Obviously, the employees who are no more than statistical figments of the Bureau of Labor Statistics "birth/death" model do not actually contribute to the output. Fake jobs may please the stock market and give politicians something to crow about. Still, they have the drawback of skewing productivity measurement. The millions of fake hires in the government's official employment counts disguise productivity growth.
Of course, it is also possible, as Gillian Tett suggests in an August 2015 piece for the Financial Times, "The Fed's Productivity Predicament," that there may be a "technological time lag." She notes that in the 1970s and 1980s, a similar swing occurred in which there was a slump after an earlier boom. I doubt it is a coincidence that these productivity stalls followed sharp increases in the price of oil. It surged from $3.00 per barrel in 1973 to t $12.00 per barrel in 1974; then, it trebled again early in this century.
If the revolution in employment proves to be even half as far-reaching as Frey and Osborne imagine, however, it presages a sweeping change in how people earn their livings and, indeed, as Marx suggested, a change in "all production and social relations." But not the type of change that most Marxists would embrace.
Whatever their political preferences, people whose thinking is indelibly imprinted with an industrial mindset — and therefore presuppose a government powerful enough to interrupt, regulate, suppress, and redistribute any outcome informed by markets—are likely to leap to the wrong conclusion. Many will suppose that the falling economic value of unskilled labor requires another costly escalation in social complexity: Redistribution on a global scale to ensure a minimum income for everyone.
Consider this from Google's Blaise Agüera y Arcas: As machine intelligence, robotics, and technological leverage in general increasingly decouple productivity from labor, we will continue to see unemployment rise even in otherwise healthy economies. The end state is one in which most forms of human endeavor are not required. In 30 years, if not sooner, we will be facing this unprecedented situation—whether heaven or hell depends on whether we're able to let go of capitalism, economic Darwinism, and the Calvinist ethics that implicitly underlie these systems. Without a change, of course, we will see mass unemployment drive a radical acceleration of the already dramatic imbalance between the very wealthy few and everyone else, leading to ugly conditions in the cities and ultimately violent uprising.
On the other hand, if we are able to set aside our Calvinism, we will realize that given the technological efficiencies we have achieved, everyone can live well, with or without a job. Capitalism, entrepreneurship, and other systems of differential wealth creation could still function on top of this horizontal base. Still, everyone must be fed and housed decently, have access to free health care and education, and be able to live a good life. I assume the nation-state will still be a relevant legal and economic construct in 30 years (though I'm not sure, as corporations or possibly other structures will complicate the picture); I guess that we will see both paths taken in different parts of the world, leading to misery and war in some, where either the benefits of accelerating technology are slow to penetrate or Darwinian economics are left unchecked.
With all due respect to Blaise Agüera y Arcas, who is an accomplished software engineer, I see the expectation that a middle-class standard of living will be handed to everyone, as a birthright, as an anachronism. More broadly, egalitarianism, or the demand for equality of economic results, has not been a universal feature of human societies and may not be predominant in the future. A quick review suggests why.
Egalitarianism in the "Garden of Eden"
Redistribution of the hunt was a common feature of primitive hunter-gatherer bands. Our most distant human ancestors devoted much effort to hunting large game animals. Sharing of the hunt must have conveyed substantial survival value by helping ensure the small group members were against starvation. Given that they had to hunt on foot, armed only with wooden weapons, the optimum strategy was to hunt no more than necessary and make the most efficient use of the captured by feeding everyone. An intimate group typically numbered about fifty, including women and children. Even with an extraordinarily large hunter-gatherer band of 100 individuals, a single mammoth could have provided more than 150 pounds of meat for each person.
Sharing was the only strategy that made sense in the face of a huge carcass of rotting meat. While our primitive ancestors dug ice cellars to preserve meat when climatic conditions permitted during ice ages, more frequently than not, they could not preserve meat for later consumption. In that circumstance, refusal to share provisions harvested in the hunt would only have encouraged overhunting. If each related nuclear individual family had found it necessary to fall their own megafauna, assuming that was physically possible, the result would have been the utter waste of a larger measure of perishable meat.
Unlike in agriculture, where more effort is typically rewarded with greater productivity, up to a point, greater effort to deplete herds of the wild game would be wasteful. Furthermore, encouraging a more energetic effort to deplete the herds would have reduced the prospects for the success of future hunts. A cooperative strategy was best suited to the survival and prosperity of the related nuclear families that composed the typical small band of hunter-gatherers.
Furthermore, moral hazard was not a problem under the conditions of hunter-gatherer life. There was little prospect that equal sharing of food from the hunt would encourage shirking by the abler hunters. Their refusal to perform would have condemned them. Their immediate families to hunger—a very different circumstance to that entailed in agriculture, where, as recorded in the book of Genesis 3:19, "By the sweat of your brow you shall eat bread, till you return to the ground, for out of it you were taken."
In Strategic Investment, I only provide an impartial viewpoint. I studied philosophy at Oxford, which the University claims helped me think clearly. Even so, you have the same ability that I have. So only believe what makes sense to you.
Book Discussion: The Sovereign Individual by James Dale Davidson