In every age the same question is asked when a band, tribe, or nation is confronted by a challenge, dilemma or catastrophe. If we examine the historical record, the evidence suggests that this question has traditionally been answered by a consensus of the elites. Though a case can be made that in the last one hundred years the opinion of the masses has gradually influenced the answer. We have always had faith in finding an answer and moving on. Will this faith endure? There is a growing sense that it will not. We are entering an era of shattered faith in finding an answer to What Should We Do that has a broad based consensus even if we restrict the decision to those within the existing elites.
History has provided a handrail to guide successive generations. We are at the end of the handrail and nothing that has gone before can prepare us from the technological changes accelerating throughout all systems, cultures and civilizations. You will say, well that’s been said before, a thousand times before, by someone in every age. And you would be right. It has been said.
Cultural skirmishes, wars, aggressions and belligerence have changed as populations scaled to billions. Elites found effective means to harness the power of the masses to maximize industrial production and to provide manpower for armies. Elites battled one another over resources and markets and trade routes. Controlling these strategic points led to a dominion over other elites. History is a record of one set of elites bowing , or pretending to bow, to another, one set of elites conspiring to betray one another—an account of elites fighting among themselves for power and authority. No victory was ever final. Over time the fate and fortunes of elites were never stable. The masses woke up to find new rulers and masters—newly constituted elites dictating who received an education, health care, jobs, benefits and security.
Democracy gave the appearance that the masses through trade union, social organizations, and elections could finally control and shape their own destiny. During the Great Depression, they had a say in answering the question: What Should We do? The welfare and benefit programs under FDR and the funding of mass education are a testament to their influence. The middle-class expansion followed, accelerating after World War II. Thomas Piketty’s Capital in the 21st Century has documented the concentration of elite wealth and income and the destruction of the post World-War II middle-class. For a short period after World War II it appeared the middle-class would act as a counterbalance to the power of the elites. That has proved to be a fantasy. New technology has accelerated the creation of a pervasive and intrusive surveillance state that has made it easier to monitor undercurrents and apply preventive measures against potential challenges to state/elite interest. Constitutional and liberal safeguards that were the first line of defense against state abuse of power have been undermined. Technology has undermined political and economic structures in the span of a few decades, and there is no indication of this process slowing down. The result is that the middle-class in America is in the process of being dismantled as an effective political, economic and social force. The working class and middle class have been divided and conquered within. Their views on What Should We do are largely irrelevant. The reason is that workers, blue collar and white collar, are becoming irrelevant in manufacture, marketing and distribution of goods and services.
Modern elites, in the private and public sectors, have access to technology that does not require mass labor to be productive and competitive. The middle-class is losing what the working classes have already lost—bargaining power to negotiate a better sharing of wealth and income. Robots manufacture consumer goods. Machine intelligence creates software and algorithms. The elites need far fewer engineers, lawyers, accountants, or architects and in the future their numbers will continue to dwindle. As Nick Bostrom’s Superintelligence, has pointed out, we have had periods of thousands of years where very little in terms of tools and technology changed. Generation and after generation of people occupied the same technological world. If you could time travel a person born in 900 to 1100 or from 1100 to 1300 they would have seen pretty much the same world. Go back in time an over long stretches of time nothing much changed whether political, social, economic or technologically.
Now consider someone who was born in 1950 who now lives in 2014; her experience of life today is qualitatively and quantitatively different from the year of her birth. Our technological world from weapons, information, computers, communications, transportation and manufacturing systems have been dramatically altered. Evgeny Morozov observed in the Guardian that algorithmic regulations are the beginning of our colonization by technologists: “[Our] smart world also presents us with an exciting political choice. If so much of our everyday behaviour is already captured, analysed and nudged, why stick with unempirical approaches to regulation? Why rely on laws when one has sensors and feedback mechanisms?”
DNA storage breakthrough allows 700 terabytes of data in a single gram. Magnetic Resonance Imaging (Miri) can now scan a single atom. A 3D printer can produce a metal handgun, car, or parts of a plane. Ultrafast low-power logic circuits from graphene by 2024. What will be her world when she celebrates her one-hundredth birthday in 2050? There is a not insignificant probability it will be a world dominated not by the traditional elites but by artificial intelligence. No one can predict the time when, the place where, or the forces that finally allow that final step to occur. It may be that enhanced human intelligence will create a new class of intelligent elites.
Not everyone agrees on the timing. Experts like Michael Jordan, one of the most respected authorities on machine learning, argues there has been too much hype and we are decades away from solving many of the hard engineering and mathematical problems. When asked when a machine will pass the Turing test, Jordan replied: “I think you will get a slow accumulation of capabilities, including in domains like speech and vision and natural language. There will probably not ever be a single moment in which we would want to say, ‘There is now a new intelligent entity in the universe.’” Jordan’s slow start view depends on human intelligence staying at the current level.
Scientists like Stephen Hu predict in the near future the likelihood of tweaking human beings’ IQ to 1000. There are approximately 10,000 gene variations (alleles) in the brain that correlate to intelligence. We are on the path to optimization of these genes to maximize our cognitive potential. Prenatal genetic engineering will change intelligence perimeters prior to birth. As impressive as being ten times smarter than the average person is, an AI at super-intelligent levels is 100,000 to a million times faster, with better memory, better retrieval and access, and self-editing and correcting, being able to alter, update and evolve its operating system as it learns. At this point, the ‘measurement’ based on IQ is a bit like using a car’s odometer to measure the speed of light. It wasn’t created with the capacity to measure that level, and any reading is meaningless. It is not unlike the measurement problem faced in quantum mechanics that makes us question the utility of what we measure in the classical non-quantum world. All of our heritage, values, culture, language, and morality have an implicit assumption—it is premised on a normal or Gaussian distribution (think Bell Curve) of human intelligence.
It doesn’t matter where you live on the planet, today you can be certain that no one in your community has an IQ of 1000. The technological accelerator that is happening as you read this essay guarantees such a person will during your life exist. What will that mean? What should we do? Destroy that Gaussian distribution by creating one, a thousand or a million such individuals, and what happens to those premises that underscore your behavior, consciousness, the way in which you co-operate with others and process reality?
The great transition we have entered, one that technology is accelerating at a rate that we can no longer control or comprehend is leading to an AI that will be super-intelligent. Nick Bostrom counsels that we need to slow down technology until we can increase our own intelligence, and that is essential to understand the nature of controls necessary to restrain such a super-intelligence. This would require a degree of co-operation, consensus, and commonly shared values that have never before been displayed among the elites. This is the irony, as the elites have finally found technological ways to marginalize the masses, an AI system by 2050 is likely to have displaced human elites and for the first time in history, there will be no longer a distinction between the elites and masses. They will share the same destiny in a world where a super-intelligent AI won’t be influenced, guided or restrained in its actions by our ethics, values, interest, goals, or morality.
What Should We Do? That question will no longer be relevant or meaningful for our species. Elites and masses will have passed the final post where the sign reads: stop and think about future generations rather than quarterly profit reports. By then it will be what should we have done? And we will likely ask AI for the answer to that question as by then we will be dependent on seeking high level answers from AI. And what will AI reply? With a neural stimulation that gives us pleasure, happiness and steers our mental activity away from contemplating our old habit of worrying over the range of answers and scenarios that always left us uncertain, confused, insecure and unhappy.
If you read one book this year, make it Bostrom’s Superintelligence. The prose can be dense, abstract, complicated with jargon—the writing isn’t a model of elegance or grace. But it gets the job done. Like an executioner’s axe it delivers a dramatic blow. Superintelligence is no literary masterpiece but it may be something more rare—a prophetic vision of an existential inflexion point on the near horizon. It is a call for us to wake up. Watch the daily acceleration on your screen and ask yourself with the technological and political elites are waking up to the existential threat. These elites with their illusion of understanding and power, with their influence and the leverage of their wealth, are about to be blindsided, along with everyone else, by technology they’ve funded and celebrated. In the case of a hard or fast take off, no one will see it coming until it is too late. But Bostrom, at the end of this powerful book, remains an optimist. He believes we still have a chance to put the brakes on technological acceleration, and give ourselves breathing room to work out a slow take off which will allows us to put in places controls over AI. Once AI has a hard take off and becomes super-intelligent, it will be too late to control or regulate it.
Bostrom lives in Oxford, and I live in Bangkok. I know his world, I shared it, and came from it, but I can’t help but wonder if Bostrom lived in my post-coup world of Thailand if his optimism about the future would still prevail. If the small probability of super-intelligent AI emerging in the next decades comes about through a hard take off, humanity will likely inhabit an alien environment, existing inside a post-human intelligence controlled world. How would we know? Having been through a number of military coups, the usual routine is to run patriotic music on every radio and TV channel. It is likely to be different with when AI sends out its message. One morning you wake up and its not marital music playing on every YouTube channel but music specifically programed to match your mood from all those choices you’ve made for years, along with carefully crafted images linked to your school, family, friends and all the memories that make you happy and reinforce your personal identity. What we should do will no longer be a question anyone will ask other to anyone other than AI.