Coming Soon: Real Androids! February 4, 2012Posted by Mary W. Matthews in Popular Culture, Random Observations, Religion & Theology.
I’ve always been fascinated by the second creation myth in the Bible. It is what is called an ætiological myth, meaning that it attempts to explain questions that its original Stone Age and Bronze Age audiences found puzzling: why is the human male virtually alone among the species in NOT having a bone in his penis? Why is the human female alone among the species in suffering labor pains? What other characteristics distinguish the human animal from all other species, and what led to our unique status?
It is that final question that interests me the most, especially since one rarely sees it discussed by scholars or exegetes. Our intelligence is of course the most obvious characteristic that sets us apart. But consider these:
- Consciousness — Consciousness is difficult to define, especially since it’s so closely intertwined with self-awareness. At its most fundamental, it means “not asleep,” but I think most of us could see a distinction between the consciousness of a human being and the consciousness of a chicken. Other qualities of consciousness might be said to include “not dreaming,” “not hallucinating,” “not drugged,” and “not drunk.”
- Self-awareness — Most higher mammals have some limited self-awareness (“I am Buddy, that is Princess”), but with few exceptions only more complex brains can pass the mirror test, that is, are capable of recognizing the image in the mirror as the viewer’s self. At some threshold that humans have passed but our pets have not, we reach an awareness of self that includes such perceptions as I am awake; I am an individual; I control what I do; I feel sensations; these are MY emotions I’m feeling; these are MY thoughts I’m thinking; I am smarter than the average bear; I am NOT a “philosophical zombie.” Self-awareness may be interdependent with consciousness.
- Imagination and curiosity — Imagination is the ability to form mental images, sensations, and concepts that are independent of whatever you are actually seeing, hearing, etc. at the time. Most animals have this ability to some extent, as any pet owner who regularly uses an electric can opener can attest, but only humans have flown to the Moon and back. I lump curiosity in with imagination because I think they’re related: “What’s new in the pantry? Could it be some cookies?”
- Moral judgment and conscience — Human beings are virtually unique among the species in our sense of right and wrong, virtue as opposed to sin or evil. We identify and condemn as “wrong” such behaviors as lying, cheating, stealing, defaming, debauching, murdering, blackmailing, etc. We feel guilty when we know or believe we have done wrong, and we do things to try to atone, or become at-one with the other(s) whom we have injured.
The second creation myth depicts the woman as the first theologian, engaging in discourse with a symbol of the Goddess and of wisdom about the nature of divinity, God’s intentions, and theodicy. The woman is curious about the fruit of the Tree of Knowledge of Good and Evil. Although death has supposedly not been invented yet, she knows what “you will die” means and wishes to avoid this consequence. She knows that the fruit is desirable because it is good to look at, nourishing, and will confer wisdom that will make her more like God. First the woman, then the man voluntarily disobeys God, whereupon they instantly acquire sin, evil, guilt, and shame. And that, children (so says the myth), is where consciousness, self-awareness, imagination, and conscience come from.
As a lifelong fan of science fiction, I urge anyone who is interested in characters like Star Trek’s Commander Data to read or view the AI-related classics, especially I, Robot (read it), 2001: A Space Odyssey (view it, especially if you can do so in an altered state), The Moon Is a Harsh Mistress (read it), Battlestar Galactica (view the entire 21st-century oeuvre); and Colossus: The Forbin Project (view the 1970 movie; reading the 1966 book requires more of a time investment than the story is worth).
These classics, and others like them, all proceed from the assumption that artificial intelligence automatically includes artificial consciousness, artificial self-awareness, artificial imagination, artificial emotions (even Data felt envy), and, except for the Asimov stories (and arguably Battlestar Galactica), no moral judgment or conscience at all.
I think it is far, far, far more likely that we will very soon develop an artificial intelligence that is smarter than we are, but that it will be a “philosophical zombie.” That is, it will be able to beat us at chess, like Deep Blue, or at Jeopardy!, like Watson; but it will never “come awake,” like Heinlein’s Mycroft; it will never try to take over the world and enslave us, like 1966’s Colossus; it will never try to murder us, like 1968’s HAL or 1978’s / 2003’s Cylons. It will not be conscious, conscious that it is a separate self, able to imagine the outcomes of different scenarios and to experience a preference for which outcome it should strive for. It will not have volition. It will never be able to imagine, plan, and execute actions that we humans would consider wrongdoing. (I say “never” because, come on, what roboticist has not heard of Asimov’s Three Laws of Robotics?)
I am also certain that we will never invent a computer that experiences emotions. Virtually every artificial intelligence that sci-fi has come up with has portrayed AIs that display arrogance, self-righteousness, judgmentalism, anger, fear, and other human emotions. “I have the right to take over the world.” “I have the right to murder someone I perceive as a danger to my mission.” “Your treatment of me as your slave makes me angry and resentful, so I will destroy you.” For decades, my question has always been, where do these machines get their HORMONES? These are machines that act as if their conscious self-awareness has been affected by adrenaline, testosterone, serotonin, oxytocin, and other hormones, alone or in combination. Why do so many clanking collections of hardware and software act as if they have PMS?
I think it is far more likely that any artificial intelligence that humans come up with will be “philosophical zombies.” They’ll have high intelligence, and perhaps imagination; but I don’t think there would be any advantage to humanity in trying to create Cylons. I want a personal assistant that will do my vacuuming for me, not one that falls in love with my husband and plots to kill me so it can marry him.
I’ve also spent decades believing that there is no way we will ever come up with someone like Commander Data — a self-contained android who for the most part is barely distinguishable from an enhanced human man. Artificial intelligence alone is problem enough, but consider that stuffed within a relatively few cubic feet are artificial bones and artificial muscles, sinews, and tendons; artificial eyes; artificial ears; artificial skin; an artificial sense of smell; and artificial senses of balance, orientation in space-time (“I am walking across the room”), acceleration, etc. Then you need at least 2.5 petabytes’ worth of artificial neurons for the AI’s intelligence, memory, conscious self-awareness, imagination, curiosity, ethical programming, etc. merely to match human capacity. (A petabyte is a million gigabytes.) Commander Data, whose processing speed, memory, and intelligence are far superior to ours, probably has at least 5 petabytes’ worth of artificial neurons.
Watson, the IBM computer and software that recently beat two human champions at Jeopardy, has 16 terabytes of RAM (about 1/160th as much as YOU do*). Watson’s 16 terabytes comprise a cluster of ninety IBM Power 750 servers (plus additional I/O, network and cluster controller nodes in 10 racks) with a total of 2880 POWER7 processor cores, not to mention the special air conditioners and so forth.
(*Remember, you use lots of RAM for things that Watson doesn’t have. Something like two-thirds of your brain is devoted to eyesight alone!)
A few days ago, I was thinking about Siri, Apple’s personal assistant software and “knowledge navigator,” when it hit me: We don’t NEED to invent a stand-alone android like Data, whose 5 petabytes (320 Watsons) of RAM fit so neatly into roughly 2.75 cubic feet of volume.
We have the Cloud. Where Siri “lives.”
Just as the Internet is vast and your smartphone is small, I see no reason why most of your new android’s “brain” shouldn’t live in the Cloud. I see no reason at all why roboticists shouldn’t concentrate on artificial bones, muscles, balance, and so forth. The actual physical body of our new android can have “hard” memory for things like sense of orientation in space and time (“I am sitting petting a pussycat”), muscle-memory, balance, and receptors for artificial senses. The motherboard, so to speak. The parts of our new android (or gynoid) that we would call intelligence, personality, memory, and so forth will live in the Cloud.
I think the real, true, Data-like android is only a few years away, and her name will be Siri Watson, or possibly Watson Siri-ous, or Mycroft Jeeves.
She won’t try to kill you. She won’t try to take over the world. She won’t fall in love with you. She won’t dream up new ideas for you. But I see no reason at all why she couldn’t be your housekeeper, your chef, your personal assistant, your butler, your gardener, your chauffeur, your masseuse, your plumber, your electrician, your carpenter, your handyman, or even your mobile entertainment unit, extruding Bose-quality speakers or a high-def video screen on demand.
Once the program is in the Cloud, it will just keep getting better. Right now Siri’s main strengths are things like restaurants, hotels, gas stations, and driving/walking directions. Soon it will be any sort of information. In the very beginning, the android “Siri-Watson” may only know how to boil water and work the microwave, but give the Cloud a year and you’ll have your very own gourmet chef. A year after that, you’ll be able to instruct Siri-Watson that you want duck á la Rachel Ray or steak á la Gordon Ramsey, and your unit of Siri-Watson will be able to consult the Cloud for recipes and techniques.
What’s going to be fun is designing what your Siri-Watson will look like. I’ll bet lots of people choose P.G. Wodehouse’s Jeeves, Batman’s Alfred, the Green Hornet’s Kato, the grandmother from that Twilight Zone episode, or Lord Peter Wimsey’s Bunter. A few may choose Merlin. A very few may even choose the android that Woody Allen portrayed in Sleeper. . . .