“When the evening is spread out against the sky...” (Photo credit Cecilia Wong, 6/30/2015 Los Angeles)
It’s actually an excellent movie, a suspense thriller about a Google-type mogul designing a robot for the famous Turing test: how do we know if a machine has acquired human-like cognition? In the film, a talented coder is chosen to be the human half of the test and helicoptered into a remote mountain bunker where the recluse mogul lives with his robotic creations. One beautiful humanoid woman is the subject for the coder to run the Turing test on. The film unfolds, juxtaposing the mogul-coder and the coder-robot interactions. Conversations go from the philosophical, ethical, to the emotional and flirtatious. Ultimately the coder questions his own humanness.
Ray Kurzweil’s prediction of that ‘singularity’ has everyone worried. Kurzweil is Google’s new head of AI (artificial intelligence), searching for ways to replicate—thus replace—the human body-brain. When that happens, when man and robot merge, when machines become conscious, well, that’s what he calls the singularity. But singularity in this context means multiple things, depending on whom you ask. I am here proposing one that is based on a simple fact of our existence—the need for food. Is anyone designing a robot that enjoys food and wine? The digestive system is possibly more complex than the brain. But if that happens, it would truly be a singularity—and one that in spirit is closer to the singularity defined by physicists and cosmologists: the point at which space-time becomes infinite.
To begin with, we don’t really know what human cognition is. The only definition we can agree on is that it is a process. But memory is a process, Google search is a process, life is a process… The exception is that in the robot, that process is predictable, written into the program. This is not true in humans —and problematic for the Turing test (if one human can’t tell whether he’s talking to a machine, another probably could). The human infant, from before birth, develops according to its DNA blueprint only to a point. Actually most of the development depends highly on the environment. In experiments in mice, the eyes and the visual cortex fail to develop properly if the mice are kept in the dark from birth. It can happen even when the mother had been kept in the dark during pregnancy (report here). This very likely occurs also in humans and true of many of our traits. In other words, each of us is a highly ‘custom made’ entity—even identical twins can vary greatly in many respects. Epigenetics studies this variety of ways that genes can be expressed or suppressed throughout life: learning and memory involve this process, as do our spring allergies—wetware is dynamic, and the DNA blueprint is never followed mechanically. Not so software.
What is machine learning?
Queen Elizabeth’s astronomer royal , Sir Martin Rees, who prefers an English churchyard to a Berkeley refrigerator to end his days, says:
“We’re witnessing a momentous speed-up in artificial intelligence (AI) – in the power of machines to learn, communicate and interact with us. Computers don’t learn like we do: they use “brute force” methods. They learn to translate from foreign languages by reading multilingual versions of, for example, millions of pages of EU documents (they never get bored). They learn to recognise dogs, cats and human faces by crunching through millions of images—not the way a baby learns.”
Note that search engines only find what people have said explicitly, not what they feel implicitly and haven’t expressed in common language, which in itself can be a limited (and potentially misleading) means of communication. If robots gain their information from reading these zillions of web pages, they access but a tiny portion of our collective conscious-unconscious. And, what about those who haven’t bothered to put their opinions or activities online? So if robots base their actions on this information, which is but a thin slice of our total human (and animal) experience in life’s 3-billion-year history (that is in some ways all encrypted in our DNA, ready to be pulled out when needed)—it is limited indeed. But it really doesn’t matter.
Does Google search extract or abstract information?
Sergey Brin was interested in “…search engines, information extraction from unstructured sources, and data mining of large text collections and scientific data…” (my italics), according to his vita on the paper in which he first proposed the Google search engine, along with Lawrence Page, while both PhD candidates at Stanford university in the 1990s. True, as it currently works, Google searches mostly abstracts that summarize research papers but can it further abstract that information into a coherent whole, as humans do?
Abstraction is considered an important human trait and the basis of our intelligence and creativity. As when a character asks in Ex Machina: “The computer can play chess, but does it know what chess is?” Children form abstract ideas from real life experiences, only later from language. Abstract paintings do not represent any specific objects but can evoke memories or suggest ideas and feelings. It is an important movement in 20th century art, even though it is a natural process that occurs in paintings and sculptures since cave times, from the earliest human artists. And also, would a computer know what beauty is? Well, let’s see.
Joe Z. Tsien a decade ago had translated the real-time coding of memory formation in neurons in the mouse hippocampus (a center of the brain for explicit memory), into the binary code of the computer. He discovered that groups of neurons, called neural cliques are organized in a categorical and hierarchical manner, from general to specific, with the general feature-coding cliques at the bottom of the pyramid (report here). So abstraction is inherent in the mammalian brain. It becomes more highly developed in humans with our larger neocortex, and of course, language.
Yes, we evolve, like we always have. But can we live forever?
Ray Kurzweil’s assertion that we can live forever actually comes with a huge caveat: he seeks “life extension” not just temporally but spatially—he speaks of a sort of spreading and extension of our body-brain’s capability, which is already happening: many of us are outfitted with life-aiding electronic implants, not to mention the smartphone that is un-detachable from every hand. When all this is incorporated into our body and each part is individually replaceable, yes, we can live forever. Meanwhile, we still need food.
Murray Shanahan, in his book Embodiment and the Inner Life: Cognition and Consciousness in the Space of Possible Minds (2010, p.65), states that the aim of AI is not to replicate living things: “metabolism is not a prerequisite…we should like to situate consciousness…within the larger picture of consciousness as it could be…artificial consciousness.” So biology is irrelevant.
Think of the advantages: a robot can get its energy supply from the sun with a solar panel and store it in a Tesla-style super battery. By contrast, biology’s way of capturing the sun’s energy is long-winded and wasteful—from nitrogen fixation to photosynthesis, ultimately growing the food that we feed on. Food (and survival) is the original source of our emotions—hunger and satisfaction (See my post here). Robots can thus dispense with all of it. Art, which comes from our emotions and implicit memory, will too be redundant.
Tsien speculates: “Could it be that 5,000 years from now, we will be able to download our minds onto computers, travel to distant worlds and live forever in the network?”
Kurzweil plans to have his body frozen should he not live to that day—more like 25 years in his estimation.
So, before it disappears forever, here is what art is like (did Eliot foresee our world a century ago?) :
Let us go then, you and I,
When the evening is spread out against the sky
Like a patient etherised upon a table; […] Oh, do not ask, ‘what is it?’ Let us go and make our visit.
In the room the women come and go
Talking of Michelangelo.
— T.S. Eliot, The Love Song of J. Alfred Prufrock (1910-15)
Deus Ex Machina!
In Greek tragedy, the plot would be resolved by God (an actor) being hoisted onto stage by some mechanical contraption, providing a neat ending to the play. No such luck with Ex Machina the film, where the humanoid robot, having succeeded in her treacherous maneuvers and committed murder, roams free in that wasteland of the human world. Will she rule?
So here is the singularity: it’s when we no longer have to compete for food and emotion is eradicated. Only then will we have a true robotic population that can explore other planets and survive on a plundered and depleted Earth, where the growing of food is no longer possible. Breaking from that food chain and being able to operate on electricity directly from the sun will be the ultimate achievement of artificial intelligence. It will be God from a machine.
It will also be the death of art.