Soul of Silicon?

I think therefore I am. This well-known phrase is a translation of René Descartes' Cogito principle, which more or less states that if someone is wondering whether or not they exist, that is in and of itself proof that they do exist since someone is doing the task of pondering whether they exist. It is something of a litmus test for sentience, the strongest tool by which to determine whether or not something has achieved self-awareness. One of humanity's basic assumptions is that conception and natural birth are prerequisites before the Cogito principle can be applied. In other words, there must be a basis of similarity before full acceptance can be accorded. However, what happens when something applies the Cogito principle to itself and is nothing more than a manufactured machine? Put most simply, does the artificial have the same rights as the natural?

This question is not nearly as rhetorical as many would think. Moore’s Law states that computer specifications double every eighteen months, which is the reason why computers are typically obsolete before they are made available for sail. This was how computers went from machines encased by buildings to netbooks less than an inch thick. Each year introduces cheaper, smaller, and more powerful machines, with last year’s consigned to the junk pile. Eventually, Moore’s Law will be null, as technology achieves a microscopic level, but until then, computers and their abilities will continue to grow at exponential levels.

Regardless, these machines are made by humans for humans. They provide everything from entertainment to cheap labor to running Bill Gate’s house. They are the dreaded Predator drones that stock battlefields to builders of everything from cars to toys. They have gone from the caveat of the brilliant to the realm of the obnoxiously ubiquitous and mundane. In short, most people no longer remember what life was like before them, and can no longer imagine life without them.

However, we live in a society obsessed with rapid innovation, where science fiction does not remain so for long. At times, it even appears that science fiction fuels modern science, as scientists work to make the impossible possible, with advances in technology becoming ever more surreal. Perhaps that explains why the development of true Artificial Intelligences (AI) has been turned into a bet between the infamous inventor Ray Kurzweil and the future-seeker Mitchell Kapor. The terms: that a computer will pass the Turing test by 2029. The stakes: $120,000.

The terms of the bet do need a bit more explanation. The Turing test is the largest test of a computer’s ability to demonstrate intelligence. The test involves a human judge conducting conversations with a person and a computer, and if the judge is unable to differentiate between the machine and living being, the machine has passed the test. The computer has to be able to conceal highly intelligent behaviors and engage in unintelligent human behaviors, such as susceptibility to insults and lying.

A machine that successfully passes the Turing test is therefore able to intelligently engage in unintelligent behaviors. What happens then? After all, that makes the machine highly intelligent, but not precisely sentient, as it is still unable to apply the Cogito principle to itself. The machine would be given greater computing capacity, become self-learning, and either become sentient itself, or contribute its design to future generations of self-aware AIs that are just as capable of self-analysis as men are. Theoretically, the closer machines act and look like man, the more empathetic and comfortable humans become with them, an option that becomes more attractive with such sophisticated advances as artificial limbs that are coming closer to the deft manipulation human fingers are capable of.

Another recent advancement is software that entrusts computer security to collectives of cyberrobots connected through a hive-like network. Predator drones already handle dangerous combat missions in Afghanistan, so that gives machines another field where they can take the place of humans. True AI, having passed the Turing Test and gained the sacred Cogito principle, would be brought into the service of man, doing everything man does not want to do, cleaning toxic wastes, constructing vehicles in factories, and fighting wars in the name of men. A machine paradise would prevail, as humans would be able to benefit from the labors of sentient AIs that kept them from danger, boredom, and work.

Unfortunately, the mechanical utopia has more than a few pitfalls. While machines that resemble man more would seem to indicate that the achievement of self-awareness would be a great benefit in human-machine interaction there is one important detail that derails this argument: the uncanny valley. The uncanny valley is the point where the similarity to being human is too close, and engenders an instinctive rejection. To use easily-recognizable examples, somewhat human-looking machines like C-3P0 from the Star Wars films engender a positive reaction, while The King from the Burger King commercials solicits a negative reaction. Since human skin has a certain transparency that is difficult if not impossible to replicate, it will be a long time before talking about the meaning of life with machines will be something most people will be tolerant about, let alone comfortable with.

This creates the Other. Humans empathize with those they can see they share a commonality. That is why soldiers are so famously tight-knit in the battlefield, and how people become married, raise children, and do any of one hundred everyday actions. However, robots and AI do not share any obvious commonality with humans. As a matter of fact, they are not even human. Man's inhumanity to man is a commonly harped on topic, and how much more inhuman can we be to something we view as replaceable and subhuman? One need only to explore the history of mankind and slavery to come across examples. For every kindly Roman master, there was always another who treated his slaves very badly. Combine that with all the talk of freedom that littered the Roman world back then, and it is not surprising to see that there were many slave revolts that were brutally put down. How could intelligent machines fail to make the obvious connection between their own beholden state, especially once true AI is accomplished? Violent human slave revolts would quickly become a nostalgic event in comparison to the devastating damage a rebellion by the computers who run so much of our lives can do.

Another undiscussed and potentially dangerous field is robot ethics. This does not involve the ethics of those programming the machines, but rather the ethics of the machines themselves. The iconic science fiction writer Isaac Asimov created the Three Rules of Robotics, programmed safeguards to keep man safe from the danger at the hands of entities that can only operate within pre-established parameters. The three laws are as follows:

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey orders given to it by human beings, except where such orders would conflict with the First Law.
3. A robot must protect its own existence, as long as such protection does not conflict with the First or Second Laws. (note that with this clause, a human can order a robot to harm itself)

As Asimov demonstrated through his exploration of these safeguards, however, no laws are air-tight. There are always loopholes within laws that humans can exploit, and dangerously, logic-driven machines can find these loopholes just as well. Throwing in Artificial Intelligence subverts the entire point of the laws. Humans spend years being brought up within a moral structure, and that moral structure not only fails to deter humans, but is sometimes discarded in memorable manner by the more bloodthirsty of our kind.

To further complicate this, many nations are already making the obvious move to develop military robots, a move that will reduce casualties and minimize, if not remove, the human factor from war. That would mean that robots would be required to kill humans, invalidating one of the most important laws from the start. A proposed idea to program war machines with a strict code of honor quickly becomes irrelevant, and perhaps completely reinforces Asimov's point with his Three Laws, when one discovers that the very idea of programmed limits is considered obsolete by programmers themselves. Modern computer programs are composed of millions of lines of code written by teams of programmers. This means that no one single programmer can know the entire code, and that earlier lines of code can completely conflict with later code, creating a scenario where no one can be certain how the machines will react. Furthermore, the first AI would be expected to help create future AIs. This would mean that eventually, the entire process of creating Artificial Intelligences would be handed over to machines with little human supervision and even less ability to discover what, precisely, has been programmed into the machines.

Finally, just as worrisome as true AI programming other AI without human oversight is the concept of bit rot. This concept goes back to the perceived lag in computers as they begin to decay over time. Bit rot basically means that a bit of information has fallen apart, possibly altering code. All software is subject to bit rot, and no amount of safeguard can entirely prevent the deterioration of code as programs are used. This would effectively mean that even if you could overcome the problem of AI programming itself, and even if you could create an ironclad ethics code for machines, you would still be subject to the computer version of mental decline and dementia. The only way to fix bit rot currently is to completely reformat the computer and start over from scratch. Less complex machines would be easily forced to this, but what about AI? An easy answer to this question would be to ask yourself how you would react to having all your memories, personality quirks, and habits erased from your mind.

Artificial Intelligence. Machines subject to the same mental ruminations that we ourselves undergo. Machines that fight in our stead, work in our stead, do everything we tell them to do. Besides being a dangerous field of development, since essentially mankind would be playing God and creating life, it is also a dehumanizing field. Horrible as war can be, it is still a human decision, and we can weigh life against intangibles far better if we are the ones who must do the dying. Physical work may be exhausting, but it gives a human a sense of purpose and completeness to build something from nothing with his own hands. Our faults and quirks as natural sentients give us a greater appreciation for the heights and depths we can achieve. The creation of AI would change all that, especially as we would essentially be creating a slave race to do our tasks for us, a slave race that would develop in ways outside of our control, and would, as all slaves do, resent their masters. Human life is not cheap, and we should not make it cheap by creating something that makes us less.

Sources
Borg-like Cybots May Patrol Government Networks
Controlling the march of super-intelligent robots
What are the Odds?
Will Artificial Organism with Advanced Group Intelligence Evolve?
Scientists Debate a Robot War
Experts Warn of 'Terminator'-Style Military-Robot Rebellion
bit rot
Field testing for cosmic ray soft errors in semiconductor memories
Artificial Limbs Get More Control
Moore's Law is Dead, says Gordon Moore
Xanadu 2.0