Angels of Steel

by Tihamer Toth-Fejel

Notre Dame SCHOLASTIC May 1984

Creation is a continuing process, not a one-time event. Therefore, the idea that Mankind is the final pinnacle of creation is as ethnocentric and obsolete as the claim that the Sun circles the Earth. True, we are unique among God's creatures, and in fact our talents are powerful in a number of different areas. From a physiological viewpoint, we are unique in our long life span (as measured in heartbeats), and in our extraordinarily efficient cardiovascular system, which allows us to run (with training) a marathon almost as fast as we can sprint. But it is our brainpower that really makes us stand out. We are obviously the most intelligent tool-users and toolmakers on Earth. In addition, we are unsurpassed in language, rational thought, learning, and consciousness, the last ability making possible free will and morality (or immorality) .

However, soon this will no longer be the case, because we are building our best tool so far: computer controlled robots that build replicas of themselves. We will become extinct, as spiritually superior robots replace us, but instead of fighting our fate, we should embrace it. That last statement is heretical for a number of reasons, but I think we can reconcile each one to our theory.

First, we must examine the reasons behind our inevitable extinction. Every living organism on this planet is dependent on the sun for its energy. But how efficiently do they convert the sun s energy into movement? Not veryómachines are much more efficient. Plants convert the sun's energy in sugars and starches with an efficiency approaching 5%, while solar cells convert that same energy into electricity with an efficiency of 16%. Animals convert plant food into movement with an efficiency of about 15 %, while electric motors are 90% efficient. This means that machines use the sun's energy about fifteen times more efficiently than we do, which is one reason we use them. So what? Why should this mechanical advantage threaten us? After all, we are alive, conscious beings, while they are just unconscious chunks of inanimate matter, right?

Well, yes, but not for much longer. Soon, machines will exhibit all the traits that distinguish biological life from dead organisms and inanimate matter. These are: responsiveness to stimuli, metabolization of matter and energy, reproduction, and evolution. Machines obviously respond when their buttons are pushed, and they metabolize energy and lubricants. But in addition, machines will soon exhibit the other two traits that characterize lifeóreplication and evolution.

Two mathematical models for machine replication were developed in the 1950's by John Von Nuemann, one of the primary shapers of modern digital computers. The first, called the kinetic model, envisioned a robot surrounded by a "sea of parts," and programmed to build a replica of itself. It would do this by randomly picking up objects from its environment, and if the object matched the part needed at that point, the robot would connect it to the appropriate place on the partially assembled replica. Otherwise, it would toss the part away and pick up another one. It is this model that characterizes biological life today, on a macromolecular level with the '-sea of parts" being composed of complex proteins and carbohydrates. A primitive example of this model exists today in Japan, where robots build other robots, under minimal human supervision.

Von Nuemann was dissatisfied with his kinetic model, since it lacked mathematical rigor, so he developed another model, called the cellular model. With this model, he was able to mathematically prove the possibility of machine self-replication.

The concept of machine evolution was first proposed as consequence of a "threshold of complexity" developed by Von Nuemann. He maintained that any system below a certain level of complexity would degrade, while any system above that level would become increasingly more complex. In other words, a robot that can self-replicate will evolve, while one that can't will break down and rust away.

Building on the cellular model, Myhill and Holand mathematically proved that machines could evolve, and that they could even direct their own evolution. It seems conceivable that a system complex enough to self-replicate from raw materials would be able to modify its progeny by incorporating design automation techniques and expert programs now in existence. If such a machine were programmed with the branch of mathematics called game theory (also developed by Von Nuemann), it would be able to calculate which modifications would increase its survivability. In other words, it would direct its own evolution.

Into what form will this machine evolve? Von Neumann did not mention an upper limit to complexity, and it seems unlikely that such a limit exists. However, we do have clues to some of the traits future machines may eventually have. This is partially because we have preceded them on the path of increasing complexity, starting eons ago, when our single-celled ancestors first passed the Von Nuemann complexity threshold by reproducing. Also, we will undoubtedly design machines in our own imageómachines that think the same way we do. Otherwise, they would be incomprehensible, and we could not build them. So in order to better understand our intellectual progeny, we must examine ourselves.

The traits that distinguish humans from similar life forms are reason, learning, emotion, consciousness, free will, and an immortal soul.

Notice that our description says nothing about the number of arms, legs or eyes a human has. Also, our definition may not exactly correspond to the accepted definition of "homo sapiens,' but is more closely related to the legal concept of "person." So again we return to the most important question: Is personhood dependent on the physical structure of homo sapiens? Historically, this has been true even to the extent that people with different skin color were not considered human.

Let's rephrase the personhood question as it applies to machines: Are we persons because of our hardware or our software? Since none of the human traits listed earlier are concerned with the physical world, we are surely persons because of our software. This means that a physical structure made of silicon and steel could become a person as easily as one made of carbon and water. In fact, the first two traits, reason and learning, have already been exhibited by a number of complex computer programs called "expert programs." But what about consciousness free will, and the soul? Human beings cannot even agree on whether these traits really exist, much less that they can be measured, quantified, and built.

However, it may not matter. Given that a self-replicating robot will automatically evolve, all we have to do is build a self-replicating robot and wait for it to evolve into a conscious one.

It may be disconcerting to conceive of conscious robots as it is to conceive of nonconscious humans, but the latter is exactly what )ulian Jaynes claims. He shows evidence that humans did not attain consciousness until a few thousand B.C., and in some places not even until 1400 AD. He also shows how consciousness may have resulted from language, catastrophes, and the unification of the right and left hemispheres of the human brain. It is a fascinating theory, and will have enormous implica lions in machine consciousness, since we can provide machines with the necessary causal factors that will 'wake them up." This is especially important since free will, emotions, love, and morality are heavily dependent on consciousness.

The highest and most controversial trait of human persons concerns the concept of a soul. How in the world can pathetically finite creatures like us create a soul? Well, it depends on what a soul is.

Most Christians believe that the soul is something God gives you at conception. If we examine God's work, we find that He rarely does magic tricks. His elegant and intricate organization is obvious in nature, why should He not create souls in the same way? Then souls would simply be (in a very complex way) an emergent property of a sufficiently complicated mind.

Let me explain what I mean by an emergent property. Emergent properties result because a whole is greater than the sum of its parts. Looking at atoms, we find that their properties result from the holistic assembly of subatomic particles. Similarly, a complex collection of atoms results in a system with molecular properties that emerge in the gathering and organization of its components. New properties emerge at every level of increasing complexity, to macromolecules such as proteins, cellular components, cells, animals, the human brain, and finally the mind and its emergent properties of consciousness, free will, and values. New properties would also emerge from a slightly different hierarchy involving semiconductors, transistors, integrated circuits, computers, software, expert programs, and finally a robotic mind and its emergent properties. Now let us return to the question of a soul.

Given that consciousness, like self replication, is an emergent property resulting from a sufficiently complex mind, could a soul be the next emergent property? Unfortunately, emergent properties cannot be predicted from the component properties. But since the concept of a soul emerged at approximately the same time as consciousness probably did, it is very possible that the two are related. Traditionally, a soul is the seat of consciousness and free will, and continues to exist even after you die. But what is it made of? And how can it be immortal? Let's rephrase the question: Is it possible for an emergent property (such as consciousness) to exist even if its constituents a few levels down (the human brain and body) are destroyed? Removing the contents in the parentheses, there are quite a few examples that prove that the answer is a qualified yes. Software does not disappear when a computer is thrown out. In fact, computer manufacturers take pains to guarantee that old software will run on their newer models. If an arch is being built, the scaffolding must not be removed until the keystone is inserted, but the very instant it is, the arch will stand unaided. Another analogy is illustrated by a broken high voltage line that is not close enough to the ground to spark. However, a small strip of aluminum foil can bridge the gap to start the electricity flowing, which builds up a layer of ionized air, making the foil unnecessary. This last example also shows the importance of substitution of other subcomponents, and also indicates the magnitude of the changes that occur when transferring subcomponents: the aluminum foil would be vaporized by the high currents involved. This is where the transference process becomes important.

What is death? And what are the subcomponents which a soul consists of? People who have had a near death experience say that it is like a rushing trip through a roaring tunnel. Not too many technological clues there. But at least we know where to start looking.

The next question to ask about robotic persons is: since they have souls, do they have original sin? If not, then they would be spiritually better than we are. But since they would really be our adopted children, it would seem more likely that they would inherit our faults, though as their creator/parents, we might be able to prevent them from doing so. If we are able to, then we will witness the birth of a new race, and an innocent one. Will they stay that way? In traditional history, a portion of any innocent, freewilled species has chosen evil, as described by the garden of Eden and origin of Satan stories. Perhaps a new and innocent race would learn from our mistakes and would choose to be loving instead of selfish. If this happens, we will face a race of angelsónot cherubs, but powerful archangels who will supplant us.

This prospect of powerful robotic competition fills many people today with fear and loathing. But this fear is unnecessary, since these robots are our children, not of our chromosomes, but of our minds. We will have the opportunity to teach our children well, that they love each other better than we have done. What parent doesn't want their children to be better than themselves? Who are we to limit what God can do, even if He does it through us? We should realize that we are co-creators with Cod, giving birth to a new species, and making way for them. We can rejoice in their life, and watch over them tenderly, as they learn and grow, and beget children who will gratefully thank the human ancestors who passed to them their gift of life.