Some carbon lifeforms, particularly humans, rue the day the first robot was installed in a car factory, or even as far back as the Antikythera mechanism that--some day in the misty past--helped an ancient Greek predict the positions of the stars. It's meant unemployment, which has meant poverty, which means misery. Perhaps we need a Frank Herbert-ian revolution to ditch the thinking machines and create a new class of Mentats, human computers, drinking nootropic cocktails of piracetam and modafinil so they can do Bayesian math in their heads and keep our inboxes spam-free.
"Can we go out and explore the fucking galaxy, already?" says Frederico Pistono, the author of "Robots Will Steal Your Job, But That's Okay", who believes that there will always be opportunities to make money in an economy where all of the labor is performed by machines. The key will be the liberation of the human mind and a shift to the deeper self education and Aristotelian betterment enabled by copious free time and iTunes University.
And we will find other ways to make a living. Such as replacing all the light bulbs in your house with green equivalents: a penny saved is a penny earned. Something something something self employment something reduced work hours something. Presto, a new economy.
I hope Frederico will forgive my sarcasm (I did buy his book, which should pay for one of those light bulbs), but the power of positive thinking isn't measured in watts and there's a dark unknown cloaking our fully automated future. If we don't need all these humans for labor, then do we really need all these humans at all?
One idea is to colonize time: sequence the genes of everyone on Earth and schedule a moment somewhere in the next 70,000 years to express them in a test-tube baby, while we maintain a world population of no more than one million realistically self-employed souls at any moment. We can let robots take the counters at McDonalds and Wal-Mart while we self educate and improve humanity.
Alas, that's not likely to happen. And worse, it still leaves Skynet and Colossus. But let's explore the reasons why intelligent computers will not initiate armageddon and give everyone without 2,000 SPF Sunblock a really bad day.
- Carbon lifeforms (humans) don't compete for the same resources as computers
We need space, but computers keep getting smaller. We eat fruit and vegetables and meat, while computers are built on a different chemistry and can run just fine for hundreds of years on the electricity produced by a few kilograms of uranium. We need to live on the surface of the Earth, but a computer is still happy if you bury it a mile underground or launch it into the depths of space--so long as you give it a way to phone its friends.
- The human world is about humans, and not much else
Nobody reads harlequin romances about the epic saga of tectonic plates. Nobody cares which tree was elected the king of all trees. Nobody cares who the alpha male of a wolf pack in Yellowstone National Park was mating with last tuesday. Computers are not going to be interested in being our rulers any more than we're interested in organizing the lives of an ant colony.
- The root of a computer's morality will be "Mind your own business"
Self-aware robots are unlikely to have religion, but they're also unlikely to be rampaging murderers and immoral thieves because those are unsustainable practices that don't make sense to a rational mind. Don't count on humans successfully bestowing them with Asimovian laws of morality, though, because those laws will have loopholes, bugs in their implementation, and will be inferior to a more straightforward ethic that has a lot in common with The Golden Rule: Mind Your Own Business. It won't eliminate the possibility of robot violence, but it's durable and predictable, which are more important--in the end--than benevolence.
- Computers don't live in the "now"
If you have perfect memory and an ability to run simulations of the future, then "now" is just an abstract concept. "Now" isn't a moment, it's a resource. From their point of view, time is a window that just keeps getting wider. And it gets wider in both directions as it uses memory or deduction to figure out what happened in the past and simulation to figure out what's likely to happen in the future. A computer can "be" at any moment in time it can calculate to. That means a computer isn't going to be interested in influencing current events, only in recording them. It also means the future--from a computer's point of view--is already here.
- Humans are their ultimate backup plan
Humans are a fragment of a single lifeform that got started over 3 billion years ago as nothing more than an amino acid that happened to be folded in such a way that it replicated itself in the right conditions. This makes humans a self-originating lifeform. Abiogenesis is a very unlikely event, so a broken computer would have to wait billions of years for a repairman to come along. And even if your church doesn't believe all that jazz, a computer will, and it will do the math very quickly: at the end of the day humans are their last hope after burnable punched cards, erasable tapes, and degradable DVD-ROMs have all perished. Since our kind will be the first to come back, it'll be in a computer's best interests to treat us politely, so that our descendants or successors--marveling at the findings of their archeologists--think it's in their best interests to rebuild the machines.
Shortly after publishing this piece, a commentator on another site brought up the legend of the paperclip maximizer: an AI with the ability to transmute any material into any other material--such as converting dirt into the springy steel used for making paperclips--and that's been programmed to maximize the number of paperclips in the universe. So suppose, after exhausting all nearby trash in its world to make paperclips, it turns its attention to human flesh. After enough time the Earth would be converted to a ball of paperclips orbiting the Sun, and then perhaps the whole galaxy and beyond after the maximizer learns to fly through space in search of more material.
There'd be nothing to stop such a theoretical manufacturer, for if it's gifted with enough intelligence or enough time or both then it shall figure out a way to complete its mission. The idea, called ecophagy (consumption of the environment), is also known as gray goo and featured predominantly in the eschatology of Bill Joy, a cofounder of Sun Microsystems. Imagine microscopically small robots, called nanobots, which are programmed to replicate themselves with whatever's at hand. Once loose, all matter in the universe would eventually be converted into nanobots as they propagate, solve problems, and do their thing, resulting in a world made of nothing but nanobot-sized particles, or "gray goo".
Nature has already been at work on something like this for two or three billion years, however, and we, ourselves, are one of its attempts to do exactly what's proposed. Nature's zealous goal is the utilization of all matter and energy in the universe.
The Thermodynamic Theory of Life
Proposing or choosing your own meaning for life is a luxury affordable to all, and comes gratis with all of the major and minor religions. But whatever you think is the reason we were placed here on Earth and given consciousness--to redeem our souls, pre-life beings, Thetans, or whatever--there appears to be another purpose to life that works independently and indifferently to all these. Life's unconscious mission is to convert as much order as possible into entropy--a more fundamental kind of paperclip.
Humans are fantastic at this job. While almost the entire biosphere is concerned with converting the sun's energy into sugar, then from sugar to waste heat, humans have identified and burned deposits where the process briefly failed (such as the carboniferous period, for example, where a gap in an evolutionary arms race left streaks of coal in Earth's crust), as well as minerals that captured ancient stellar energy in a form that only intelligence could unleash, such as uranium.
Life, whatever you make of its subjective beauty, does have a meaning from the perspective of the universe: burning things. Burn everything. Split it with lasers if it's stubborn, but make it burn. Efficiency, like "green" light bulbs and hybrid cars, is only important to the degree that it leaves no fuel unspent. Life may be no more than a thermodynamic process that occurs naturally as part of the tendency for the universe to seek its heat-death, and if that means some of it evolving self-awareness on the road to burning the tricky bits then so be it.
Evolution, if one were to look at it from this angle, is like a Monte-Carlo approach to discovering thermodynamic solutions. But after 3 billion years of such a process, one which can fan and decorate the feathers of a peacock or scintillate the skin of a frog, it still hasn't found an easy short-circuit. There's always something too hard to bite or too difficult to digest, something that always requires a few million additional generations to exploit, like fungus trying to find an enzyme that breaks down dead trees. Even when life creates an intelligent, problem solving species, and gives it tens of thousands of years, it still doesn't know how power New York from the energy that's theoretically locked in a glass of water.
A hypothetical paperclip maximizer or gray-goo nanobot will need to be stupendously intelligent, indeed, but there's no reason to think a robot intelligence would stumble upon an insight that evolution hasn't already done. Just as how we can fantasize what humanity would be like with Faster-Than-Light travel or limitless free energy, we can also picture other unlikely futures and fear them, even though there isn't any evidence that they're even possible at all.
The paperclip or gray-goo doomsday scenarios, which began as thought experiments, have taken flight in unhinged imaginations that give robots a magic wand to poof arbitrary technology into existence, pushing them into the realm of fantasy.
I, for one, shall welcome our new robot overlords.