Futurologists tend to look at the impacts of artificial intelligence, genetic engineering, nanotechnology and robotics in isolation. It is a simple matter to single out a technology and then make a linear regression projecting its future. The most commonly referenced current example is ‘Moore’s law’. Gordon Moore was originally speaking about IC complexity – which was applied by Dave House to computer performance, predicting it doubling every 18 months. Popular acknowledgment of Moore and House is that ‘Moore’s law’ has successfully predicted the exponential growth of digital electronics over the past 45 years. Ray Kurzweil’s book The Singularity Is Near: When Humans Transcend Biology has a theory of technology evolution that predicts the same exponential growth for modern economic and biological systems. This religious belief in technology’s beneficial growth has been termed “Apocalyptic AI” by Robert Geraci, assistant professor of religious studies at Manhattan College in his book of the same name. He suggests that Apocalyptic AI is a technological faith that derived from Judaism and Christianity. It certainly takes religious faith to make predictions of technology’s impact on humanity so detailed that Ray Kurzweil expects near term technology advances to enable his immortality. Unfortunately Ray’s fervor to outlast a normal lifespan has eclipsed his more impressive insights about the three overlapping revolutions of genetic engineering, nanotechnology and robotics. I agree with Ray Kurzweil’s postulate that machine intelligence, nanotechnology and organic biology will merge together to create synergistic results. It is this merger of these three disparate scientific disciplines that is thrusting humanity into a new age.
Prof. Dr. Hugo de Garis, recently retired from his role of Director of the Artificial Brain Lab (ABL) at Xiamen University, China. Ten years ago Hugo headed a team in Belgium, Europe tasked with designing the world’s first artificial brain. At the time, he was concerned that he was building the first massively intelligent machine ‘artilect’. He predicted that artilects would become infinitely smarter than human beings. In 2005 he wrote a book titled: The Artilect War: Cosmists vs. Terrans : A Bitter Controversy Concerning Whether Humanity Should Build Godlike Massively Intelligent Machines. Prof. Dr. Hugo de Garis’s willingness to continue his work in China building artilects is an example of the ethical dichotomy frequently seen in technologists. They commonly feel comfortable developing potentially dangerous science as long as they have made society aware of the impending ethical choices. In counterpoint, this is the same dilemma that motivated Bill Joy to change his career in 2003, leaving his fabulously successful computer company, Sun Microsystems.
A more moderate view on the danger of artilects can be found in Jeff Hawkins recent book On Intelligence. Jeff Hawkins has spent his life studying both the human brain and computer science. His perspective on artificial intelligence is grounded in both biology and computers. Jeff Hawkins’ book describes a comprehensive theory of how our brains work and then uses his theory to design computer intelligence emulating human thought processes. He has termed his theory Hierarchical Temporal Memory (HTM). Jeff Hawkins has founded a company, Numenta Inc., to develop computer technologies that replicate the structural and algorithmic properties of the neocortex.
HTM offers the promise of building machines that approach or exceed human level performance for many cognitive tasks. Just a few years into development; HTM is already enabling low cost, generic, learning based systems with particular success at emulating human vision as applied to image analysis and object tracking. In the next chapter I will propose one dramatic potential of an HTM like computer which combines all elements of genetic engineering, nanotechnology and robotics. In the chapter The Future Of Intelligence, Jeff Hawkins proposes many fantastic, previously inconceivable applications of HTM in the next few decades. He counters the fears of Prof. Dr. Hugo de Garis’ artilects causing social disruption. Jeff Hawkins questions whether intelligent machines will exhibit any characteristics capable of the ‘Nightfall’ scenarios quoted from Ian Morris in my chapter Predestination. His opinion is that intelligent machines are going to be one of the least dangerous, most beneficial technologies we have ever developed. I agree with his observation that there will be no reason to include the animal derived old brain instincts like fear, paranoia, and desire. Intelligent machines have no need for an analog of human emotions. Instead they will be dedicated and optimized for tasks like vehicle navigation or weather prediction.
As we move into the latter half of this century, brainlike machines will surpass all higher order human thinking capabilities. These artilects will have speeds a million times faster and capacities trillions of times larger than individual human brains. If you consider that the Internet only reached a critical mass in the early 1990s; in the last 20 years it has become one gigantic memory store for humanity’s knowledge. With the geometric growth it will continue to experience, its eventual capacity is beyond human imagination.
Intelligent machines designed to tap into the global knowledge stores that supersede our today’s Internet will have capabilities not imagined. How humanity is likely to interact with these vast computer intelligences is the topic of the next two chapters.
MAN/MACHINE
Computer scientist J. Storrs Hall’s book on the development of artificial intelligence and its ethical aftermath, Beyond AI: Creating the Conscience of the Machine, says the following about our likelihood of proceeding rapidly down the path of merging man with machines:
“We’ve been modifying one another since the invention of speech, enhancing our memories since the invention of writing, enhancing our eyes with lenses, our muscles with machines, and our voices with electrical signals on wires. It is so ingrained in human nature for us to enhance ourselves that ‘not’ doing it would be the weird, inhuman thing to do.”
By the middle of this century we will be further supplementing our bodies with genetic, drug and electronic augmentations. Many futurists including J. Storrs expect inexpensive robots to be common in this time period. I’m in general agreement with this projection. My major caveat is that these will not be robot butlers and general purpose assembly workers with human like behavior. I believe that these robots will be purpose built appliances seamlessly integrated into our lives. A natural evolution of all the technology we take for granted today like cell phones, personal computers and high speed transit. Think for a moment how many electronic devices in your life fail within four years, are obsolete in two, and require more power than what they replace. This characteristic of the creations of the industrial revolution has not changed. In many ways the useful life of a device has shortened while the energy and materials required to design, build, and maintain them has increased. Nano technology and its power for self-assembly and self-repair has the potential to change all this. I think it will, but not in our current century. Consider that the human brain works all day on a couple of bananas, yet our current super computers require more than 50,000 times the same energy input. Consider that technology is getting so complex it is not profitable to repair or upgrade. Consider the constant battle for control of our electronic devices against cyber-attack; the attacks modifying as fast as biological viruses. Our defenses are always tardy and frequently ineffectual. Consider that a humanoid robot butler will not only have to have a superior general purpose computer to emulate a human, but must also have a power source to operate untethered, and must have physical properties that simulate the muscles and joints of the human body. It must operate for extended periods without failure of any critical component. It will have thousands of moving components susceptible to wear, and environmental degradation from moisture and grit.
Compare this to a biological butler; the result of many millions of years evolution optimizing its systems for operation up to periods of one hundred years. Humans use comparatively little energy, we are self-learning, self-repairing, adaptive, and creative. There are already six billion of us on the planet. There is no way that humanoid robots will be cost effective in this century. The biological alternative is orders of magnitude more efficient use of resources. What will happen instead is the augmentation of humans. Consider that with synergistic application of genetic modification, Nano technology, customized drug delivery systems, and brain-to-AI linkage we will have humans targeted for particular societal functions. The economics of this alternative are so dramatic that humanoid robots will not be competitive except for unique environments like radioactive areas or outer space. The primary roadblock to progress in augmenting humans is ethical. Storrs Hall’s ruminations on this topic are erudite and extensive. Our ethical systems evolved to get human life to our current epoch. The most significant hurdle facing the human race is how to blend current philosophies developed over tens of thousands of years with incompatible ethics appropriate to future humans. The religious confrontations bedeviling society will be greatly exacerbated by augmented humans. It will be like creating millions of sociopaths. There won’t even be commonality within the groups of augmented humans. Being modified purposefully, they will characterize new emerging social mores. Simply put, they will think differently. We don’t have any analog of how people connected from birth in a hive mind will think. The hive mind will share advanced computer artificial intelligence. A hive seems very likely to engender new religions and atypically human ethical patterns.
The Pew Internet&American Life Project statistics showed that in 2010 half of teens send 50 or more text messages a day, or 1,500 texts a month, and one in three send more than 100 texts a day, or more than 3,000 texts a month. Two billion of the earth’s population or one out of three people alive use the Internet. It is this generation that is going to be making decisions about using electronic implants to connect themselves to the artificial intelligence systems of the future. As J. Storrs is quoted at the beginning of this chapter, it would be atypical if large numbers of people from various cultures didn’t enhance themselves. Once enhanced these people will immediately have extended intelligence, be more connected, and more productive than non-enhanced humans. The rudimentary beginning of the hive mind will be non-threatening. Just extension of what people are already doing on a frequent basis. As the biological connections mature to handle higher bandwidth and multiple senses; as the artificial intelligence of external connected computers improves; as the relations between connected individual deepens into new communication patterns we will discover that society has diverged. The young fully embracing these sweeping changes and the older generations resistant and further alienated.
ANDROIDS/BIOBOTS
As research work progresses on new frameworks of intelligence like Jeff Hawkins’ Hierarchical Temporal Memory (HTM) we will soon begin to see artificial intelligence with specific capabilities exceeding humans. The Deep Blue chess computer that used massively parallel brute force computing to evaluate 200 million positions per second when playing chess with Gary Kasparov; did not exhibit learning or creativity. New artificial intelligence designs will be self-learning, able to creatively recognize patterns on top of patterns. These artificial intelligences will learn from experience. Early applications will be autonomous vehicles, medical diagnosis, and weather prediction. None of these uses exemplify Ray Kurzweil’s three overlapping revolutions of genetic engineering, nanotechnology and robotics. I’m going to present what I consider a likely scenario that will be possible within fifteen years. This scenario will also lay the groundwork enabling the first generations of future humans.
Human genome DNA sequencing cost has steadily fallen. Within ten years there will be millions of fully sequenced people. In another decade whole populations will be sequenced, particularly in progressive homogeneous locations like Iceland. Initial uses of this data have been genetic screening and secure identification. Consider that this is the code of life. Unraveling that code will yield fundamental understanding of all life’s processes. At this stage in our biological sciences this understanding is beyond our collective grasp. I believe it is within the grasp of an artilect.
What follows is a likely outcome starting in about 15 years. The enabling technology will be brainlike memory systems based on Hierarchical Temporal Memory (HTM), computers custom designed to support the HTM model, the lowered cost of human genome sequencing, and emerging within-body nano-sensor technology.
Imagine a new breed of artilect (I call it Genome-Brain). The primary data input for this self-teaching Genome-Brain is our genetic code. Our genome provides a natural pattern for the hierarchic self-learning input. A second sense (sense, in the model of sight-hearing-touch) would be less structured data. It would consist of medical records, intelligence tests, aptitude tests, education records, occupation history…associated with individual genomes. A third sense; instead of being fixed DNA patterns or static personal data records would be real-time biology data transmitted from nano-sensors within the bloodstream of a few thousand volunteers. The Human Metabolome Project has already identified 8,500 metabolites in humans. These cellular by-products will trace every chemical generated by cell metabolism in real-time. The Metabolome provides the biomarkers to associate gene expression with human bio-chemistry.
According to HTM theory, human creativity results from the mixing and matching of patterns in our neocortex. The cortical hierarchy is multi-level. Creativity is a result of abstraction of relationships of higher-order objects; recognition of patterns on top of patterns. After the ‘Genome-Brain’ has spent a decade accumulating billions of relationships between our genetic code, our real-time biology, and our resultant behavior, it will be able to identify specific groups of genes that control measurable traits like human intelligence. It will also be able to relate real-time biology with disease and genetic susceptibility. More interesting for the human race will be its capacity to begin to design new genetic material intended to modify the genetics of individual humans. Craig Venter’s company Synthetic Genomics has already synthetically created the genome of a bacterium from scratch. In twenty-five years it will be common place technology to modify human DNA. The tough part will be to know what to modify. Genome-Brain will provide custom designed life forms to organizations like Synthetic Genomics. Feedback from the initial genetic designs (monitored by permanent nano-biosensors) will enable the Genome-Brain to refine its knowledge and enable improved genetic designs. Concurrently technologies for human computer implants and custom drug enhancements will complement these new genetic designs. For example, rudimentary computer interface brain implants will be co-evolved with genetic changes that enable improved communication between neurons and electronic brain implants.
Within fifty years the first examples of Technology Augmented Bioengineered (TAB) humans will be living among us. I’ve coined this acronym TAB to represent an intermediate form of human life, predecessors to the androids or biobots created towards the end of the decade. Unfortunately we humans will not be prepared for the social and ethical consequences. Prof. Dr. Hugo de Garis’ warnings about the Artilect War are grounded in humanity’s inability to rapidly evolve its social mores. After all, we are genetically programmed over two million years to enable our existing social structures. It is quite likely that the instincts enabling the growth of society were intrinsic to our success as a species. The disparity between genetically modified augmented humans will create massive social strife. The advantages of redesigned life will be so persuasive that some groups will ignore the consequences. Be it longer life, increased intelligence, or merging into the hive mind; the new humans will be so superior that jealousy and fear are the guaranteed reaction of the unenhanced multitude. This has been a popular topic for science fiction. It is no longer fiction. This will happen during our lifespans.
Comments