British artificial intelligence researcher expects in a few decades time robots will become highly desirable partners in relationships.
Levy is an expert in artificial intelligence. He is fascinated with the idea of "love and sex with robots," and his visions of the future include "malebots" and "fembots" as lovers and life partners. A chess champion and the president of the International Computer Games Association, Levy, 62, has just published a book, "Love and Sex with Robots: The Evolution of Human-Robot Relationships" -- that is provocative in the truest sense of the word. He is convinced that human beings will be having sex with robots one day. They will show us sexual practices that we hadn't even imagined existed. We will love them and respect them, and we will entrust them with our most intimate secrets. All of this, says Levy, will be a reality in hardly more than 40 years from now.
I can understand how relationships with artificial intelligences could be very satisfying. An artificial intelligence could be programmed to have similar values and interests and to be very patient and engaging. But I do not expect robots to become sexually attractive unless materials for robot construction become so advanced that robots can look very like humans.
But will robots find us attractive as companions?
"The mere concept of an artificial partner, husband, wife, friend or lover is one that, for most people at the start of the 21st century, challenges their notion of relationships," says Levy. "But my thesis is this: Robots will be hugely attractive to humans as companions because of their many talents, senses and capabilities." Given rapid developments in technology, Levy believes that it is only a matter of time before machines will be capable of offering human-like traits. According to Levy, "love and sex with robots on a grand scale are inevitable."
Also, will robots be autonomous or will they be part of large collective minds? Will the machines moving around have their own self-contained artificial intelligences and autonomy? Or will they be so closely connected into massive stationary AI computers that there won't be all that many distinct separate artificial intelligences to have relationships with?
I'm thinking humans will find their ideal mates by genetically engineering them.
Our brains have an inertial reference unit (and FuturePundit reacts by thinking it would be great to have a far better one bioengineered into us).
Researchers have discovered a sophisticated neural computer, buried deep in the cerebellum, that performs inertial navigation calculations to figure out a person’s movement through space.
These calculations are no mean feat, emphasized the researchers. The vestibular system in the inner ear provides the primary source of input to the brain about the body’s movement and orientation in space. However, the vestibular sensors in the inner ear yield information about head position only. Also, the vestibular system’s detection of head acceleration cannot distinguish between the effect of movement and that of gravitational force.
Dora Angelaki and colleagues published their findings in the June 21, 2007 issue of the journal Neuron, published by Cell Press.
But if this inertial reference unit (IRU) place in the brain was really great then it could integrate a history of motion for a long time and tell you how far you've travelled on a trip just like aircraft IRUs can. So this discovery makes me think some scientists have found the location of a brain subsystem that is in need of a great deal of design improvement.
Can we come up with micro-miniature gyroscopes that can out-perform the semicircular canal of the inner ear in detecting motion and rotation?
Angelaki and colleagues based their brain studies on the predictions of a theoretical mathematical model postulating that the brain could compute inertial motion by combining rotational signals from the semicircular canal in the inner ear with gravity signals.
We need better gyros and better accelerometers. Plus, we need a better integrating biocomputer that constantly calculates one's location and velocity.
Watching an episode of that tired prequel Star Trek: Enterprise where genetically engineered Suliban fought Jonathan Archer in temporal war got me thinking about genetic engineering for decreased susceptibility to death from accident or murder. The Suliban had rather unusual skin that seemed like it might form a protective plating. I can imagine some humans deciding to sacrifice their looks for a reduced risk of death.
So here's what I want to know: 50 or 100 years from now (assuming the robots or nanobots do not take over and that we don't all just live in virtual reality) will people choose to genetically engineer themselves and their offspring to have biological protective plating and other enhancements to make them harder to kill? You might argue that most people won't do that. But won't some people do it?
Some forms of protection won't have any esthetic costs and so should be pretty popular. For example, blood-borne nanodevices that can store many hours of oxygen will make people much less at risk from drowning or smoke inhalation. Also, people will go for nanobots that can seal off arteries and veins severed in accidents or shootings. One can imagine nanobots that can serve as distributed blood pumps so that a shot to the heart couldn't kill you either. But after getting all the invisible protective enhancements some people will desire even more security and even greater defenses against sudden death.
So how about biologically constructed nanosheet body armor skin made by your own cells? Surely enzymes that can construct carbon nanotubes are a pretty small jump from enzymes that can create carbon polymer lipids for fat storage.
Body armor nanosheet skin coating will come with some downsides such as less attractive visual appearances and less soft touch. Unless other humans get their brains reengineered to get a sexual turn-on from such skin anyone who walks around with unnatural-looking skin will find their sexual options much more limited. But not everyone has a strong sex drive and some will choose to use a gene therapy to turn down their sex drive even lower so that their reduced physical attractiveness won't leave them feeling frustrated.
Silk worms can synthesis incredibly strong silk and other organisms make other powerful structural materials. The goal of creating skin cells that create armor seems quite attainable. Powerful computer simulations will help guide bioengineers in their search of the solution space to produce designs that can create incredibly strong materials. All the naturally occurring biological materials will serve as starting points. So the question comes down to esthetics. How much will people trade off looks in order to reduce their odds of sudden death?
Although mice, like most mammals, typically view the world with a limited color palette – similar to what some people with red-green color blindness see – scientists have now transformed their vision by introducing a single human gene into a mouse chromosome. The human gene codes for a light sensor that mice do not normally possess, and its insertion allowed the mice to distinguish colors as never before.
In a study published in the March 23, 2007, issue of the journal Science, Howard Hughes Medical Institute researchers at Johns Hopkins, together with researchers at the University of California at Santa Barbara, demonstrated in a series of cleverly designed color vision tests that the genetic modification allows mice to see and distinguish among a broader spectrum of light waves. The experiments were designed to determine whether the brains of the genetically altered mice could efficiently process sensory information from the new photoreceptors in their eyes. Among mammals, this more complex type of color vision has only been observed in primates, and therefore the brains of mice did not need to evolve to make these discriminations.
The new abilities of the genetically engineered mice indicate that the mammalian brain possesses a flexibility that permits a nearly instantaneous upgrade in the complexity of color vision, say the study’s senior authors, Gerald Jacobs and Jeremy Nathans.
While this study used a human gene in mice I think it has implications for human upgrades. Imagine a future in which some people genetically engineer their offspring to see a much wider range of colors. Instead of eye cone cells with 3 absorption peaks for light frequencies imagine 10 different variations on cone cells with 10 different peak absorption frequencies for light.
Or how about hearing? Imagine a human with the ability to hear up into much higher frequencies of sound. We could use genes from dogs or other mammals to enhance us for greater ability to distinguish sounds just as these scientists created mice with a human gene to give them greater ability to see colors.
Retinas of primates such as humans and monkeys are unique among mammals in that they have three visual receptors that absorb short (blue), medium (green) and long (red) wavelengths of light. Mice, like other mammals, only have two; one for short and one for medium wavelengths.
Nathans suggests that these knock-in mice mimic how our earliest primate ancestors acquired trichromatic vision, color vision based on three receptors. At some point in the past, random mutations created a variant of one receptor gene, located on the X chromosome, producing two different receptor types. Present-day New World (South American) monkeys still use this system, which means that in these monkeys only certain females can acquire trichromatic color vision.
In contrast, among Old World (African) primates such as humans, the two different X chromosome genes duplicated so that each X chromosome now carries the genes for both receptor types, giving both males and females trichromatic color vision.
Imagine you were going to genetically engineer the perfect dog (I expect people will some day). Well, wouldn't you want to give him greater ability to see colors?
In the comments section of a previous post, "Human Stem Cells Partially Fix Rodent Stroke Disability", Brett Bellmore opined:
I sometimes in my darker moments suspect that the sigularity will consist of lab rats taking over the universe, due to regulation delaying human trials of all the various treatments and augmentations.
Brett's idea is that if human enhancement is so restricted by government regulations then under relatively less restrictive regulations animals could receive genetic enhancements that would make them intellectually so improved (in science fiction writer David Brin's terminology "uplifted") that they'd surpass humans in intellectual ability. While he was jesting this does not seem impossible.
A more likely outcome in my view is that some countries will allow much more offspring genetic engineering than other countries and their populations will far surpass the rest of humanity in intellectual ability.
What I find hard to predict is just which countries will have more or less restrictive approaches toward genetic engineering of offspring for the purpose of cognitive enhancement. Current Western left-liberal denial of genetic causes of differences in IQ won't last for another 10 years once cheaper DNA sequencing leads to the discovery of lots of genetic variations responsible for differences in intelligence. At that point will leftists come to see genetic engineering as the best way to end poverty? Will right wing hawks decide that genetic enhancement is needed to compete militarily?
One might expect the Chinese to embrace genetic enhancement without ethical qualms of the sort that show up in discussions in Western countries. On the other hand, the Chinese leadership might decide cognitive enhancement is bound to produce too many people who are so smart that they will lack sufficient deference to authority (ie they will oppose the regime). Genetic enhancement might come to be seen as a threat to stability and a threat to those in power.
Offspring genetic engineering will probably come with a high initial price tag. However, that doesn't mean that offspring genetic engineering will be out of reach of poor people in industrialized countries. Elites might decide (quite correctly in my view) that government subsidies for genetic enhancement will pay off many times over for governments that subsidize it. 160 IQ children who might overwise have been born with 80 or 90 IQs will be orders of magnitude more productive and hence will generate enough tax revenue for governments that any costs for the genetic engineering will get paid back many times over.
On the Marginal Revolution blog Alex Tabarrok and Tyler Cowen are debating transhumanism. Tyler discusses how much will people be willing to genetically engineer their children when doing so makes the children be less like their parents.
Most people want their children to look like themselves, and to some extent to think like themselves. We invest many thousands of dollars and many months of our time to acculturate our children. Now let's say your children could be one percent happier throughout their lives, but this would mean they were totally unlike you, the parent. In fact your children would be turned into highly intelligent velociraptors and flown to another planet to live among their own kind. How many of us would choose this option? I can think of a few responses:
1. Transhumanism will bring improvements of more than one percent; we should forget about identity and let everyone become healthier and happier. What's wrong with uploads?
2. Governments should not restrict transhumanist innovation. Let people and their children choose their degrees of identity continuity for themselves. (Isn't there a collective action problem here? Everyone wants a more competitive kid but at the end humanity is very different.)
As for the idea of making kids 1% happier: It will become trivially easier to genetically engineer offspring to feel happier: give them genes that make their minds feel happy even in the face of adversity. I do not think that large physical changes in shape or other non-cognitive body changes will be needed to make happier people. I do not even think that IQ boosting will be needed to do that. My guess is that genetic engineering for happiness will be aimed at directly enhancing the feeling of happiness independent of other characteristics that might also be changed in offspring.
Regarding uploads: the ability to copy one's mind into a synthetic brain will cause some severe problems. One can easily imagine a sort of arms race between different identities competing for influence. Some will try to acquire wealth in order to get the computing capacity needed to create many copies of their brains. Competition for resources will probably become much greater when copies of sentient entities can be made quickly.
Uploads are also problematic because a person could be copied against their will and then the copy could be modified to be more compliant and willing to work for some cause that the original mind would reject as immoral. Imagine a great weapons scientist kidnapped and copied in order that some country or group could have many copies of an extremely talented scientist to work for their cause. The world would become a much more dangerous place.
Even without uploads the potential exists to some day infect a person (or an entire population) with a virus that changes personality and motivation. This capability would be attractive to governments, business interests, and people in relationships. Not sure if he loves you? Genetically reengineer him to make him more committed to long term relationships.
Transhumanism will never make as large a difference between a single generation as does immigration.
Fortunately, change across a single generation is likely to be small so parents will say yes even though 5 or 6 or 10 generations down the line the changes will be dramatic. It's because of this wedge effect that Fukuyama is so worried about relatively small changes today and it's precisely for this reason that his opposition has no hope of success in a free society.
I disagree with Alex regarding the potential rate of change across generations. The rate of biotechnological advance is going to accelerate by orders of magnitude because biotechnology will increasingly be driven by the same technologies that drive computer technology advances. For example, microfluidics devices will be fabricated using some of the same technologies used to create semiconductors and microfluidics devices in all likelihood will have lots of semiconductor gates and analog electronic circuits built into them.
We will have all the existing genetic variations in humans to choose for offspring. But we will also have the genetic variations for similar proteins in thousands of other species to investigate to look for variations that yield some desired quality. Plus, as we come to understand the genetic signalling system regulating cell growth, differentiation, and other functions of the cell lots of obvious ways to modify genes to create desired effects will jump out at us.
My guess is that in 30 or 40 years time a person planning to have an offspring will be able to choose from millions and perhaps even tens of millions of well understood and functionally significant genetic variations. Personality type, assorted behavioral tendencies, intelligence, and many physical characteristics that determine abilities and esthetic qualities will be selectable. Combinations of genetic alleles to code for physical appearances and cognitive characteristics that have yet to naturally occur in humans will be available to put in offspring. Download "Alligator Boy".
The ability of people to introduce huge number of genetic changes from one generation to the next is not the biggest reason to be worried about what sorts of semihumans or transhumans might be created (though that will be problematic). I think the real problem with big changes in sentient beings (either through genetic enhancement to create transhumans or human-machine interfaces that create cyborgs or uplifting other species - see David Brin's science fiction novel Startide Rising and the sequels) is the potential to create intelligent minds that do not have some of the emotional and ethical structures that cause human societies to function.
For example, the tendency to dole out altruistic punishment could some day be genetically engineered out of offspring. Think about people who report crimes they witnessed being perpetrated against others or who take the time as witnesses to step forward and volunteer to testify in a criminal trial. Imagine that the genetic variations that code for the rewards brains deliver to themselves for doing acts of altruistic punishment were just edited out when some people designed the genomes for their offspring. Well, that'd lead to a decrease in the rate of criminals being caught and of punishment of non-criminal abuses of people.
So work backwards to transhumanism. We cannot and should not ban it, but to what extent should we regulate/tax/patent/subsidize it? We don't know until we determine the value of identity at the margin. The Icelanders -- all 279,000 of them -- are not crazy to insist on some language skills for their immigrants; they would otherwise be overwhelmed and lose their way of life. The fact that their customs are changing each generation anyway -- and often quite radically -- is beside the point. Nor is it relevant that many Icelanders emigrate to the U.S.
Imagine, for example, the first genetically engineered generation is made to be, on average, so averse to killiing animals that they all oppose fox hunts. This just happened in England without the use of genetic engineering to produce the inter-generational change in attitudes. With genetic engineering we can expect successive generations to have far greater changes in values than we are witnessing in the West due to effects of industrialization.
When the demand for a change in personal identity is strong it can have important external effects. You may not want to be a velociraptor but if I change what choice do you have? Or you may simply have a preference (atavistic and irrational perhaps but still a preference) for human beings as they are now.
Tyler makes the mistake, however, of jumping from such and such preferences are important and real to such and such preferences justify regulation/taxation/subsidization etc.
Certainly some day the awareness of some parents that other parents are starting to genetically enhance the intelligence or motivation of their offspring will cause many in the first group of parents to follow the lead of the second group simply in order to keep their own kids competitive.
Transhumanism is going to present a problem to libertarians: On the one hand libertarians will tend to favor a laissez faire approach to regulation of offspring genetic engineering. On the other hand genetic engineering will easily be able to produce offspring that like to follow orders and that dislike those who are not like them in some way. If some fraction of society decides to use genetic engineering to produce offspring that are more communist or more totalitarian in attitude or extremely religious and hostile to non-believers or criminal then libertarians are going to have to decide whether highly coercive government intervention in the short run is worth tolerating in order to prevent far larger rights violations in the longer run.