Can Machines Have Souls?
I mentioned a couple of weeks ago that 2018 is the fiftieth anniversary of Stanley Kubrick’s great film, 2001: a Space Odyssey, which stars a computer named Hal who is able to think for himself and ultimately kills almost all the humans for whom he is supposed to be caring. Hal is not the first dangerous thinking machine in fiction nor is he the last.
But the field of so-called Artificial Intelligence, AI, is booming today, and we have made great strides in creating computer algorithms which can talk to us, can decipher our speech and translate it from one human language to another, which can recognize faces and make some sense of a visual field, can diagnose diseases and which may soon be driving our cars.
Now when I use that word algorithm, I mean simply a procedure. Computers work by codes, detailed sets of instructions. If this is the case, then do that, otherwise do something else. If you were to print out the operating system of your computer or any computer-like device, you would have a long series of ones and zeroes that would make no sense to you but would make perfect sense to the computer.
The nearest analogy to a computer operating system in our own bodies is our DNA, but you know that your genetic inheritance is only a small part of what makes you you.
AI is all around us and is a very hot topic on the news. And the organization in which I have been active twenty years, the Institute on Religion in an Age of Science, or IRAS, has chosen artificial intelligence as the topic of its summer conference in June of 2018, and I have been honored to be selected as the chaplain at that conference, which means that I’ll be preaching on the subject for six morning chapel services. If any of you is interested in attending this year, please talk to me. It’s a beautiful place and a very special experience. Just ask Frank or his wife Barbara, IRAS’s President.
It’s an apt topic for that organization, for Artificial Intelligence is right at the crux of intersection of religion and science.
What would it mean to have a machine which was conscious? Would such a machine be able to see deeper into reality than human minds could? Would it be wiser? Would such a mind be subject to human shortcomings such as rage, cruelty, indifference, depression, racism? Would we be committing murder if we turned off the computer which generated that mind?
Those are real questions, but I have framed the big question here in a way that might make it harder to get a precise answer. I have said the question is whether machines can have souls, but that question only has meaning if we know what a soul is. And souls are notoriously hard to define. The whole effort to define words, to parse them so that we can know precisely what we are talking about, comes up short when we come to the word “soul.” All of our Zen wisdom, all of our existentialist and paradox-loving and transcendentalist instincts, not to mention all the popular songs of the last fifty years, arise to say, “hey, if you gotta define it, you don’t understand it. Soul is just something you have.” But of course, that attitude makes the word almost useless for a searching scientific or philosophical inquiry.
Some people would say that soul is the most basic thing there is. Emerson, for example. In the reading we did earlier, the soul exists beyond cause and effect: “As there is no screen or ceiling between our heads and the infinite heavens, so there is no bar or wall in the soul where we, the effect, cease, and God, the cause, begins.” The individual soul is a manifestation of God and God is the oversoul, the sum of all souls.
Yet in the religion and science dialogue, the word “soul” is so resistant to analysis that it might be said to hinder, rather than help, understanding.
`When I announced on Facebook that I was preaching on the topic, Can Machines Have Souls?,” one of my retired UU minister colleagues commented, “no. no. no. no. no.” From the religious side there is a strong argument that nothing created by humans can have the special human quality that we call “soul,” whatever it may be. Let me sketch some of the religious ideas that support this argument.
In the Garden of Eden story in Genesis, God forms the first man, Adam, out of the clay and then breathes into him the breath of life. The clay is non-living substance, but it is God who imbues it with life by his breath. This is similar to Greek mythology, where the Titan Prometheus creates humans out of clay and blows the breath of life into them. That much seems to have been OK in Greek myth, but then Prometheus thinks the humans ought to be able to cook and keep warm, so he steals the element of fire from the Gods and gives it to humans; Zeus gets angry at this theft and sentences Prometheus to be chained to a rock where ravens attack him and devour his liver every day.
In both the Greek and Jewish perspectives, you don’t want humans to get divine attributes. In Genesis there is the story of the Tower of Babel, where humans were going to reach to the heavens, so God created different languages for them to speak so they couldn’t understand one another and couldn’t coordinate the construction. It’s not good for humans to aspire to be Gods.
Fast forward to the fourth century Council of Nicea, when Christianity had been adopted as the religion of the Roman Empire, and Constantine called all the bishops together to determine what Christianity was. Theologians of the time were pretty sure that God the Father had existed from all time, but they weren’t too sure about the second person of the trinity, the Logos which had been incarnated in Jesus of Nazareth during his life on earth. Had the logos existed from all time, or was it created by God the Father at or before the time He created the rest of the Universe? This question divided the Arians, who said that the logos came later, from the Athanasian, who said the logos was coeternal with God the Father. At the Council, the Athanasians won out, and produced the Nicene Creed which is still said in most orthodox Christian churches. One version of that creed says that Jesus was “begotten of his father before all worlds,” another that he was “begotten, not made.”
What does that mean? In the old Biblical terminology, “begetting” is what a male does to produce a child, while the female “bears” the child. To “make” something is to form material into a shape, and the thing that a human makes will be non-living. Even a sculptor who forms clay into a human form will have only created a likeness. To make a human, from the male point of view, you beget, you father. The Nicene council insisted that the logos proceeded from God the Father as a child is the product of his father, not as a sculpture is the product of the sculptor’s art.
This Judeo-Christian idea that a human can’t make a soul is carried into stories like Pygmalion, the sculptor who falls in love with his statue, Pinnochio, the woodcarver who wishes his puppet into life, and the sorcerer’s apprentice, where the lazy assistant animates a broom to do his task for him and then doesn’t know how to stop it. All of these have unhappy endings, and the moral they teach is that trying to treat a made thing as alive is bound to result in disaster.
In Jewish lore, there is a creature called the Golem, which was fashioned from dust or clay but animated using religious rituals and made to perform various tasks; in most versions, the Golem could not speak. Usually somebody got hurt dealing with the Golem.
In the nineteenth, Mary Shelley published her novel Frankenstein, or The Modern Prometheus. This was both one of the first horror novels and one of the first pieces of science fiction. We often call the creature Frankenstein, but properly Frankenstein was the name of the creator. This creature not only could speak and read, but reason. It had the ability to murder people and to entertain motives such as revenge and rage. It was designed to be scary and it was.
What I’m suggesting here is that the Judeo-Christian religious tradition and a strain of lore all concur that beings made by humans, as opposed to those begotten by humans, were not supposed to be animated or have souls. At the same time, the Industrial Revolution was producing more and more machines to take the place of humans in agriculture, industry and the home. In the twentieth century, machines were applied to thinking tasks as well as manual labor.
The word “robot” was invented by a Czech playwright, Karel Capek, in his 1921 play R.U.R. The initials of the title stand for Rossum’s Universal Robots, which was the name of the company featured in the play, a factory which is making lot of money manufacturing workers for a large variety of enterprises around the world. These creatures are not composed of metal or silicon, but of flesh and blood, yet they are made in a factory, not born in a natural way. A plant assembles them just like a Ford assembly line. Though they look about like humans and can talk and interact, they are treated by the management as things. When one of them breaks, it is simply sent back on to the factory floor to be disassembled and its parts used in other robots.
The conflict in the play is set in motion when a liberal reformer arrives insisting that these robots must be accorded human rights. As the plot develops, the company management decides to make the robots more human-like by giving them the capacity to feel pain. This pushes them into a new kind of consciousness and eventually they revolt and start killing the humans. So R.U.R., the drama which introduces the concept and word robot, also introduces the dread of a robot takeover.
Okay, that’s a thumbnail sketch of the religious side of the question, how about the scientific? I am just getting into the reading on this subject, but I know that developments of the last twenty years in neuroscience raise profound questions about how what we call consciousness arises in the brain, how our brain generates our minds. We know a lot more than we used to about how the brain works, but we still don’t know how it generates consciousness. The Enlightenment in the Eighteenth Century gave us many great things – modern science originated there – but it also overemphasized reason as opposed to emotion. The ordinary consciousness and intelligence that we know of is embodied intelligence, and it is tied in to human emotions at a basic level. Thus Descartes was wrong to say, I think, therefore I am. It would be more accurate to say I feel, therefore I am.
As you sit here right now, the feeling of being you is not just your thoughts. You may be thinking, you may even be thinking in complete sentences, but you also might be turning over several streams of thought at once, listening to me drone on, wondering what you’re going to have for lunch, replaying the argument you had with your significant other last night or with your mother three decades ago. There may be a soup of words floating around in your mind. And part of the mix might be an itch on your right ankle, a queasiness in your gut, an ache in your back. All of that is part of what it is like to be you at this very moment in time. And it’s a set of things that you obviously can’t program in advance into a computer.
What we call Artificial Intelligence is not a program per se; it is a means of teaching a computer to program itself, the way the brain does. Our brains evolve by a kind of neural Darwinism; there are a million billion connections in our brain, and they can fire in circuits. When we are babies, the circuits are firing at random, but soon it becomes clear that some circuits get us what we want and others don’t. The circuits that win for us get reinforced and the circuits that lose get discarded.
So we can teach machines to program themselves the same way. This is called deep learning.
How will we know if we have succeeded? Some of you may be familiar with the name Alan Turing. Turing was a brilliant British computer scientists who helped the Allies crack the German Enigma codes in the Second World War. In 1953, he proposed a test for artificial intelligence. Can a machine can fool you into thinking you’re dealing with another a human? If so, we have achieved artificial intelligence.
Can deep learning produce emotion? Such as a poet might tap in composing a poem, or a composer might tap in composing a piece of music?
We did a Turing test just now, when I asked you whether the poems in the trifold were written by a human or machine. Let me ask you briefly about the first poem, My Beautiful Enthusiasm. How many people think that was written by a real person? And How many think it was written by a machine? How about the second poem, Down In the Bass? Real person? Machine? Ok, here’s the answers. The first poem was actually written by a machine programmed by Alan Turing himself. The second was written by Langston Hughes.
An interesting twist on the Turing test was posed by the 2013 movie, “Her.” By now we are used to computers, phones and other devices that incorporate a person-like program to interact with the user. If I want to find a piece of music on my iPhone, I can just press the big button at the bottom and say “Siri, play something by Ella Fitzgerald.” That saves me having to scroll through a bunch of menus to get what I want. Many of us have those home speakers which respond to the name Alexa.
Well, the movie “Her” takes this one step further. It is set in the now or not-too-distant future, and involves a guy named Theodore who makes his living ghost-writing love-letters for other people, but whose marriage has come apart and he has a hard time sustaining real love with anyone. He gets a new computer which comes with an operating system, the basic program that operates a computer. But this operating system has artificial intelligence. So as soon as he turns it on and chooses a female voice for it, the operating system names herself Samantha and riffles through the data stored on Theodore’s computer to learn all about him in seconds. Then Samantha starts interacting with Theodore in a very personal way, teasing, cajoling, pouting. This is not computer as servant, assistant; this is computer as best friend. Of course it turns out badly. Poor Theodore, madly in love, is crushed when he finally comes to realize that Samantha is on intimate terms not only with him but with 500 other users at the same time. He has always known she’s not human, but this drives home the point that she isn’t exclusively his.
That’s “Her,” the movie – some clever and imaginative Hollywood writing. I found it stretched my credulity that anyone could fall in love with a being without a body. But I think it teaches something very valuable about what it is to be human. To be human is to be embodied. Our minds are carried by our brains which are wired in to every other part of our bodies. In the movie, Samantha yearns for a body but without one, a romance with an embodied human is really not sustainable.
I don’t know what a soul would be without a body, so I don’t know whether a soul can be immortal. By the same token, I have a hard time imagining God, a being with an intelligence and will but without a body. We are programmed by our instincts to see and recognize faces even when there are no faces; we see faces in clouds and trees and rocks, we can make faces out of clay. We recognize signs of personhood in inanimate objects because our brains have evolved social instincts. We project personhood on the world and call it God.
The Dilbert cartoon on front of the Order of service may seem inappropriate to this discussion, for the intern Asok is a real person, though he says he has given his soul over to social media. I included it because it seems to point towards a real trend. We have not created machines which can replace our individual brains, yet, and it may not be possible to do so in the future for a variety of technical reasons. But we are already in a strong interaction with machines; we use machines to connect us to other human beings.
Emerson thought, as I said earlier, that God was the Oversoul, that God consisted of the sum of all souls. Well that Oversoul at this point definitely has digital components. We are interacting on any kind of digital media by the algorithms which govern that media. I am a user of Facebook, which has just changed its basic algorithms twice in the last two weeks, and I see that I am getting a significantly different slice of reality in my feed. This will have ripple effects even on those of us who dp not use a particular corner of social media.
What is the bottom line here? I don’t think we have reached the point where digital devices or their algorithms are going to substitute for real people in a deep way. While we may find our personal digital assistants useful in finding a pizza parlor, the next song we want, or keeping our calendar, we are not likely to ask them out for dinner and a movie or confide our deepest hopes and fears to them. We may fold them into our entertainment strategy, but it would be a big mistake to look to them for emotional or spiritual sustenance. They are not your new best friend. But they will play an ever-increasing role in our connections with other flesh-and-blood people.
Readings, from the website “Bot or Not?”
Here are two poems, either of which might have been written by a computer or a flesh-and-blood poet. Which do you think wrote each one?
1. My Beautiful Enthusiasm
YOU ARE MY BEAUTIFUL ENTHUSIASM. MY FELLOW FEELING
SIGHS FOR YOUR SYMPATHETIC TENDERNESS. MY TENDER DEVOTION IMPATIENTLY PINES FOR YOUR YEARNING.
MY BURNING INFATUATION LOVES YOUR ADORATION. MY LOVELY CHARM SEDUCTIVELY CARES FOR YOUR LOVABLE
2. Down in the bass
That steady beat
Walking walking walking
Like marching feet.
Down in the bass
They easy roll,
Rolling like I like it
In my soul.
Riffs, smears, breaks.
Hey, Lawdy Mama!
Do you hear what I said?
Easy like I rock it
In my bed!