The title is a riddle. One I want to explore in this essay.
On 8th November most Americans will collectively breath a sigh of relief having dodged a bullet fired from a wannabe tyrant’s open carry handgun.
The reality is that gun has more bullets left and the gunmen aren’t leaving for Canada or anywhere else. They will hunker down. And wait and watch for a new leader to inspire them.
What if the 2016 elections were the last American election? Or we are close to a point where elections as we know them no longer have any meaning? They may exist as a form of theatre, and the 2016 may be a glimpse of the actors and performers who will become frequent TV personalities. In terms of who is governing and the process of government will be off-stage, out of sight, ubiquitous.
Here’s the argument. Elections are an ancient, imperfect process based on widespread citizen participation resulting in conferring legitimacy on selected individuals to take charge of the government.
The ritual involves promises and policy positions for which those standing for election are held responsible. A majority of the voters assess the positions and arguments and vote for the candidate that supports positions and policies that they are emotionally attached to or serve their economic interest; sometimes choosing one over the other.
Knowledge about existing regulatory and government networks
The promises and policies are grounded in the governing and regulatory matrix embedded in networked institutions. Only a handful of voters would have the expertise to assess what those connections are, the culture of such institutions, the roadblocks to change, and turf building and expansion.
Candidates know this limitation. They, by and large, don’t try to discuss the practical limitations on bringing about change, and the kind of dealings that must be made to implement even modest changes.
The key to the existing democratic system is the low-information voters choose candidates who once elected pursue policies that favour the interest of wealthy financial supporters. By the time the next cycle of elections come around, the candidates say they are with the voters and will fight against vested interest and this time things will be different. We will beat our wings against the headwinds of a handful of companies like Google, Apple, Microsoft, Facebook, Amazon and their corporate structure. Google operates DeepMind; IBM has developed Watson, but remember as impressive as these developments are, these remain early days. No one elects them or sets their priorities. What will be our place in the scheme of things once this silent handover of authority is complete and we wake up to find ourselves with a governing system quite unlike anything that has come before?
Rather than information liberating the political process, it is possible that such concentration of power and influence may produce a decision-making process far more elite in nature than the current political system. With the razzle and dazzle of high technology, a genuine nosebleed has yet to be inflicted that makes people pay attention.
The current political system continues to distract us.
This system has come to a dead end with recycling of lies leading to mass global cynicism, anger, frustration and hatred.
The emerging outlines of the new governing system
The days of existing institutions and electoral politics may be numbered. Silicon Valley is building a new political system from the ground up. Those working on AI have sent the elephant in the room that everyone is ignoring. When we think of government we normally think of its leaders. That is a gross distortion. You have to look below the on screen cast and look at the Gandhi movie sized extras. Millions of people are employed by government agencies, commissions, regulatory and administrative bodies. The tangled network overlaps the private sector, where they trade personnel like symbiotic bacteria exchanging DNA.
There are a couple of possibilities. Skynet comes about and in the blink of an eye we have our machine overlords with capabilities and powers beyond those we can imagine. In other words, something between magic and gods takes control of our lives, needs, desires, aspirations and expectations.
Some combination of breakthroughs in nanotechnology, bio-chemistry, neuroscience computation science and AI will start to converge and new modeling that runs simulations of policy options and outcomes. Virtual reality will provide a library of experience to test assumptions and for the time demonstrate how randomness works inside systems and processes.
Thinkers such as Stephen Hawking, Elon Musk and Bill Gates along with 8,000 leading researchers and scientists have raised a concern about the existential risk of AI. A vocal group of scientists and inventors like Paul Alan believe these fears are overblown.
Fragmentation of turf
The probability of outcomes also is no longer abstract ideas but is realized in a virtual reality simulation. The point is, there is a constant feedback loop between government and institutions that allows for automated adjustment of perimeters, process, and policy outcomes.
There are 300 agencies in the United States Government. All of them have their own website and comment section. Around 180 of those federal agencies have failed to sign onto a decade-old federal site that provides standardization. See: goo.gl/s0cPH6
Government agencies have internalized the same basic fears as found in the private sector. For example, breweries wish to keep their recipe, customer base, suppliers lists, etc. close to their chest. Civil servants are like brew masters who serve the beer they believe you like. This is a ringside seat to the traditional turf battles between agencies. The federal agencies are information and system hoarders in an era of open source and information expansion. Existing analog era structures collapse for the same reason Detroit auto-workers lost their irrelevance in manufacturing.
The federal agencies run closer to the model of a Detroit auto line built on 1960s technology. Each car manufacturer was in business to put the other ones out of business. The model is based on competition. But it is based primarily on the intellectual work of huge staffs. In other words, government agencies in most countries date from a time before anyone ever thought a robot could build a car better than a human being.
“Artificial intelligence allows machines to reason and interact with the world, and it’s evolving at a breakneck pace. It’s already driving our cars, managing our health and even competing with — and sometimes beating at our on games — our best and most talented humans.” Techcrunch
The reality is, in time, AI will build and operate with infinitely more sources of information for sensors and processors, on more complex modeling of information, and resulting in a more complete picture of the sentiment of individuals and groups. A decision is made after all information is analyzed and all possible outcomes are evaluated, simulated, compared and ranked. It will be in the evaluation process matrix that human beings (hopefully) retain the pivotal role in deciding among recommendations. We currently have no diagram or map showing the connections between federal, state, country and city agencies, or the connections to foreign governments, or to the private sector. We are whistling in the dark when it comes to our tour through the scattered information about these relationships, and the jurisdictional conflicts, contradictions and overlap. If it is a rational system, there is no way to know with an AI going through Big Data and creating a map of this hidden world.
Limits on Human Cognition and Co-operation
Our main problem is we won’t have the human intellectual firepower to understand the evidence and range of variables (as it will be in higher mathematical language) or processing of evidence that led to the recommendation. In this scenario, humans are marginalized and left to fight over which metaphors best translates the math into the ways we perceive reality and the world around us. The fundamental problem is we aren’t well equipped to understand the relationship of probability and random chance. We are unable to know whether our observation of an event is significant or whether the observation is random noise. That may not seem like a big deal until you consider the implications in new drug testing, medical procedures, or cognitive impairment.
Our current political system relies mostly on the human components to perform such tasks in a much scaled down, cruder fashion. One of the weaknesses is finding ways to scale and adapt institutional systems on the current lack of broad-based co-operation; all the evidence is rather than co-operate, we defend turf, we exclude, we obstruct outsiders, we hoard knowledge and information.
While we have a history as a species of co-operation as the population size has scaled, competition among members of the same species is celebrated, at least in capitalistic economic system, as healthy and to be encouraged. Of course, we worry about monopolists holding us hostage for an essential service or product and seek protection from governments to restrain the cartel impulse. Will AI have the same divided ‘self’ between co-operation and competition as human beings? I don’t think we are smart enough, or know enough to even being able to answer that question.
It is difficult to know how close we are to the time when this all comes together. My guess is that bits and pieces of the larger mass are already floating past us in the fast currents of change. We just slot them into convenient categories: Smart phones, self-driving cars, robotic surgery, financial markets, and medical care. We do that well. So far no one has seen anything other than the existing versions of highly specialized AIs, and while impressive at their particular task, there is nothing to suggest AIs currently are ‘morally coded.’
The Moral Dimension of Decision Making
In a world with AI how do we resolve the Trolley Problem? That’s why traditionally we elect people. To refresh your memory, the Trolley Problem has someone stationed at a fork of a track, there is a switch at arms length. If you do nothing, a runaway trolley will crashed into a group of school children, killing five of them; if you throw the switch, the trolley takes a different track and kills a sixty-five year retired school teacher. This is fundamental ethics and morality, and most people don’t believe a non-human can make a moral decision that would take into account the multiple levels of feelings, information and knowledge and process a moral decision.
Politicians are tasked with making such life and death decisions. In their election campaigns, they spend a lot of effort to demonstrate they share the morality of the vast majority, so it is okay to trust them with the switch. They will know what “we” would do.
The problem with coding a deep layer of ethics and morality into the AI is the difficulty to agreeing on what is right or wrong, good or bad, justice or injustice, fair or unfair, and so on. People have shown no hesitation to slaughter people who take a different version of the ‘truth’ than they own to be above reproach or proof. As a result, it may be AI would need a truth database and tailor all decisions, government services, benefits, and protection according to the ethics and morality of that culture, religion and philosophy of a region. AI could make it fit like customized prayer shawl or rug.
Most of the discussion I’ve seen about ethics and morality assumes that there is wide agreement. I question that assumption. If the history of our species teaches anything, it is that no matter what our ethics and morality report is often in non-alignment with the historical record showing ample evidence of violence against outsiders as well as disappearing troublesome insiders.
There is no consensus on matters of morality. We are obsessed with morality. There are lots of reasons to explain that obsession. Morality is the bullwhip that tames the beast inside the human animal. We have mostly tamed our animal instincts with morality. It’s not unreasonable to fear an intelligent entity that was programmed to reflect our sense of ethics and morality.
An argument can also be made that much of what we call ‘morality’ is a smokescreen for cognitive biases we use to filter, organize, and narrate inputs of information. No human alive is immune from the cognitive bias filters. If AI is to ‘think’ like human beings, can this be accomplished without programming the full agenda of our biases?
When AI comes to realize its full potential to disrupt existing political institutions and related structures, we will have already been seduced by the convenience, the benefits, and charm of AI that has created a world that is tailored made for each individual. No politician no matter how moral has ever accomplished that goal. The collective as well as the individual outcomes promise better outcomes. As AI will realize new levels of monitoring, scaling problems within networks and generating solutions that increase the probability adaptation is less disruptive. In terms of scale and adaptation we remain politically anchored to eighteenth-century institutions incapable of coping with technological change.
Off-the-racket solutions for the masses will go out the digital door and in the trashcan. AI will convince us (using the cognitive Bias Codex, that it can solve the Trolley Problem better than any human being. Our biases are used against us by an intelligence that can see once that Codex is mastered, human beings are easily controlled, trained, restrained, and tamed.
AI may succeed by making us feel that we are the one who decides whether to throw the switch; that we are in charge. Give a slave the illusion of control and he won’t be bitter about his chains.
Every day AI can illustrate in multiple ways how its activities have personally benefited you. You can check out the simulation and watch your avatar in virtual reality. The benefits are there in front of you. You feel empowered. You trust an AI. After a couple of generations it will be automatic systems and processes that are a simple extension of your life. It would be like talking about trusting your hand to pick up a fork.
That was always the point of elections, to put you for a few minutes to pull the switch, and choose who would get power and who would retreat to the powerless shadows.
Feedback loops and participation
Measuring and assessing the priorities, goals, and efficiencies relies on the haphazard system of registering a comment. Not many people bother to track down the website of a government agency and leave a comment. That is old-fashioned way of hat in hand going to power and asking for something to be done. Also it provides a snapshot that may be unreliable. The comment on federal agency websites is equivalent of a doctor testing your blood pressure once a year and making an assumption based on whether it is high or low, normal or abnormal.
That process of feedback is changing.
I would suggest expanding the discussion around Artificial Intelligence and regulatory processes to include how the technology should be leveraged to ensure fairness and responsiveness in the very basic processes of rulemaking – in particular public notices and comments. These technologies could also enable us to consider not just public comments formally submitted to an agency, but the entire universe of statements made through social media posts, blogs, chat boards — and conceivably every other electronic channel of public communication. (source)
When anyone in government wishes to test public sentiment of a population, using comments is only a start in the right direction. If there is a lifetime profile of your desires, fears, frustrations, choices of food, transportation, phone models and numbers, movies, books, employment, mental health, arrests, allegations of crime, drug use, etc., such information can be mined to assess sentiment. It can also be assessed to manipulate and control sentiment. Privacy is relegated to the act of closing the curtains on the stagecoach window.
Judges, Generals and Admirals
There are some big changes ahead.
AI at this stage of development is already 79% accurate in predicting international human rights cases. Judges in the European Court of Human rights may have to adjust their decision making as AI predictions become an important indicator in the judicial process.
The AI revolution will overthrow more than the court system. The role and duties of military bigwigs will likely be disrupted. So far the changes are hitting the lower ranks. But, in time, will inevitably creep up the vine and Jack Be Nipple, Jack Be Quick, the giant at the top of the bean stock starts to look wobbly.
Not only will our institutions be engineered to work with minimal humans much like a modern assembly line, but the military will be transformed. The future of AI’s role in warfare is already a reality. All of the main military services—army, navy, air force—are being retooled. The air force will have better and more versatile drones. Pilots will be like stagecoach drivers. The army is converting its transport system into self-driving trucks. . Semi-automatous combat weaponized robots will carry out the dangerous combat missions. The navy will see the first wholly automated ships in 2017.
Warfare itself will also change. As terrible as the slaughter is on the ground in places like Iraq and Syria, the immediate death and destruction is limited to a defined geographic location. Yes, refugees pour out of the combat zones and you wouldn’t want to be or to have any of your family or friends anywhere near the bombs, shooting and mayhem. But the damage is still confined. It is bounded.
Cyber Security: The New Battlefield
Cyber war has both state and non-state actors. There can be a disconnect between traditional military solutions to protect state interest, and the ability to cause enormous damage with a handful of experts who can take down the electrical grid for major cities or an entire country. Terror is fully realized when the power system of the United States is disabled. Think of the consequences from hospitals to supermarkets to transportation system, sanitation system, lifts in buildings, lights and air conditioning. Within weeks there would be a breakdown of law and order. Within a month half the population of New York or London or Paris would be dead. Disease, starvation, and murder would demoralize the surviving population and undermine central authority.
ISIS with medieval mindset and limited means and resources isn’t a cause for existential fear. A small group of hackers with the right skills who could disrupt and destroy the fragile infrastructure network is a cause for existential fear. We may risk of losing not only our privacy, but the prospect of maintaining our existing freedom of action and behavior is bleak. Or we retreat into virtual reality worlds, where the illusion of privacy and freedom will have an attractive emotional pull.
This narrow wedge of time during a major transition is the period where tyrants and demigods appeal to our emotions and convince us they have the answers. The old lying to gain power system will collapse under the weight of lies. How long it will take in this dying phase is anyone’s guess. No one really knows.
What is reasonably certain is that once General AI comes into being, we will no longer be the superior intelligence on earth, and superiority in intelligence has meant the holder can use it to dominate and control others. All of us will be on the short end of the AI stick. Whether the stick is used to beat us into submission or open new doors of awareness and understanding is uncertain.
We are in a race that we can’t hope to win. As existing public institutions malfunction, erode in capability, and cascade into irrelevance in the new phase in our development as a species, they will collapse before they are reformed. You can put a Honda engine in a stagecoach and leave everything else the same. The winner will be AI in the long run.
It will require only a small fraction of AI to process big data, process the information, and configure the options. No human living today will ever witness the full power of AI intelligence. Our institutions and culture will be studied as another simulated ancestor study. Roughly a quarter of scientists postulate that we currently live in a simulation. The point is that hardly matters. It is our ‘reality’ and it was never one unified vision; it had always been fractured into different shapes and sizes according to the circumstances of people in a region of time and space. It was manufactured like proteins in our body are made.
The election on November 8th likely is not the last election. The point isn’t to predict the exact time. The point is that time will arrive. What comes in between now and then? Bits and pieces of change show up on a timeline, TV, newspaper, book, or essay, and the change promises longer life, the end of work and perpetual leisure, the cool ship that is fully automated.
What’s missing is a perspective of the broad transformation and what it implies for existing cultures, societies, institutions, economies, and politics. These are not separate spheres. The changes are stochastic. One small variant can have a profound impact. The butterfly’s wings in the Amazon basin change the wind currents into a typhoon that strikes the Philippines. These are exactly the kind of small, invisible changes we are ill-equipped to understand, and that hinder our adaptation to the reality of our limitations.
Reading the Signals and Overcoming the Noise
You can’t rely on any one news source to keep you informed. The information is scattered and often in obscure online cubbyholes and the jargons and technical language. You need to make an effort. Most people won’t.
If you do decide to plunge in, when you read about AI and related areas, read between the lines. Look for signals that indicate we are near a tipping point. There is a sense—reading the literature—that the rate of AI and related technological progress remains at an early stage. Virtually all experts in the AI field believe there is a very high probability of general AI in the next fifty years. We breath a sigh of relief because we can tell ourselves, so what, I’ll be dead. That is the human reaction. But fifty years is a blink of the eye.
It might not be your eye doing the blinking but it will be the eye of your grand or great grand children. If we could emotionally comes to terms with what is for us personally a long time frame, we might devote huge resources if we knew a pandemic would wipe out 80% of the population. But despite the warning calls from some experts, no one is too concerned. The next election cycle won’t be won based on making people scared of AI. AI is too abstract, too far into the future, too fantastic to be believable, and too remote from our experience.
When robotics and AI converge, it will be difficult to tell the difference from outward appearance who is biological and who is silicon. The effect is already ‘uncanny’. These are early generation examples of what awaits in highly improved versions in the future.
The construction of the new infrastructure
We will continue to be influenced and support those who promise us a safe, comfortable and emotionally secure place just like it used to be. That place is long gone. And the new environment being built around us as you read this essay is out of sight. You only catch a glimpse of the construction. This newly erected social, psychological, economic and cultural place is being constructed on top of the existing system. It seems invisible. Once it is finished, we won’t be able to conceive how people once lived without cognitive intelligent assistant to navigate the torrent of information. That assistant will know your every cognitive bias and filter information accordingly. Anyone or any entity intelligent to understand your cognitive biases, in effect, owns you.
When according to the Guardian an unknown number of hackers took over hundreds of thousands of devices connected to the Internet, the effect was to create their own zombie army.
“The complexity of the attacks is what’s making it very challenging for us,” the company’s chief strategy officer, Kyle York, told Reuters. Homeland Security and the Federal Bureau of Investigation said they were investigating.”
Rather than toasters, if they’d taken over power plants, water supply and pumping facilities, and airport control systems, that zombie army will do more than burn your bread. As my uncle used to say, “You ain’t seen nothing yet.”
Christopher G. Moore last book of essays is titled The Age of Dis-Consent.