Singularity describes the merging of human and computer intelligence and the rise of super-intelligence as a result. Proponents of the idea of singularity try to posit it as the next step in human progression, where humans will cease to exist as currently constructed and will instead transcend our given form and become a hybrid race that is part computer, part human. Singularity has been portrayed in popular culture in several movies, the most popular of which are the Terminator and Matrix movies.
Vernor Vinge, a science fiction writer, first wrote about the vision of technological singularity and coined the term in 1993. He wrote, "Within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended."
Ray Kurzweil, inventor and futurist, is a fervid proponent of technological singularity. Kurzweil predicts the timeline of singularity as follows:
In 2011, Ray Kurzweil sponsored a movie/documentary about singularity, titled "Transcendent Man," which has been screened in five major cities in the U.S., as well as London. In December 2012, Kurzweil was hired by Google as a director of engineering to "work on new projects involving machine learning and language processing."
In 2000, Bill Joy, a well known computer scientist and the primary figure behind the BSD operating system (on which MacOS was built on) and the widely used Java programming language, joined this discussion. In a Wired magazine article, "Why the future doesn't need us," Joy declared, in what some have described as a "neo-Luddite" position, that he was convinced that growing advances in genetic engineering and nanotechnology would pose severe risks to humanity.
Proponents of singularity often cite Moore's law to support their claim. Moore's law states, crudely, that the capacity of computer chips doubles every two years. That is, the speed and capability of computers grows at an exponential speed. Such an exponential growth is a powerful enabler. Consider the series 1,2,4,8,16,32... The small increments in the beginning may be misleading about the overall speed of the series’ growth. The 20th element in this series would be 1 million. The 266th element in this series is 1080, which is more than the number of atoms in the universe.
Proponents of singularity argue that thanks to this exponential growth, the processing powers of computers will reach such high levels in the next few decades that it will be possible to simulate the human brain in high fidelity. The workings of each neuron in the brain will be simulated in real time, achieving a full simulation of the brain. At that point, the computer will essentially have the equivalent of human intelligence. In the succeeding years, with the increase in capacity, the computer intelligence will be several folds ahead of human intelligence.
Opponents of the feasibility of singularity cite that exponential growth is hard to sustain. Exponential growth is seen in the beginning of a series, but then due to limitations/adversities, most series will level off and stay constant. An example of this is the population of rabbits. Initially, the increase is exponential; however, due to scarcity of food sources and an abundance of predators, the population stabilizes around a constant. Therefore, opponents of singularity argue that the exponential progress of computer processing speeds will similarly hit a brick wall. At the chip level, physical issues such as heating will make exponential speedup unsustainable. At the cluster level, latency, consistency, and scalability issues will also prevent exponential growth.
Underlying all of Kurzweil's ideas regarding the progress of technology and the singularity is the Law of Accelerating Returns. This Law states that technological progress occurs exponentially instead of linearly, meaning that each new advancement enables several higher advancements instead of just one higher advancement, and, concordantly, every year brings more useful inventions and discoveries than were made in the last. The first generation artificial intelligence (AI) approaches failed, but simulating a human brain may work if we know the workings of the brain in excruciating detail. As a promising development, recently, “deep learning” and “deep neural networks” technologies achieved great success in image and speech recognition tasks.
However, the opponents of singularity like to point out that the workings of the brain as a whole are still a big mystery. We have information about the rough mechanism of how a neuron works. An excited neuron can transmit a signal to a neighboring neuron through its synapses. But, there is no clear explanation about how thought occurs from this process. Brain-scanning techniques are improving, as they are based on computers, but the brain may throw us more complex surprises as we learn more about it.
In fact, much of the brain power comes about through organic materials, and the very low-level analog physical interactions between these materials. These physical phenomena could be close to impossible to model/simulate in a digital environment. Henry Markram, lead researcher of the "Blue Brain Project" for simulating mammal brains at the molecular level, has stated that "it is not [their] goal to build an intelligent neural network." He claimed, “[That would] be very difficult because, in the brain, every molecule is a powerful computer and we would need to simulate the structure and function of trillions upon trillions of molecules as well as all the rules that govern how they interact. You would literally need computers that are trillions of times bigger and faster than anything existing today."
Another relevant question is whether we can develop the parallel processing architectures needed to support the parallel processing that goes on in the brain. The brain uses far more parallel processing than exists in most classical computing designs.
Even if a computer successfully simulates the human brain, whether such a computer design will be “scalable” to two times, ten times, or even one hundred times the brain’s normal power is an unknown; for the human brain’s computation power may be inherently unscalable. Also, if a computer models the human brain, human emotions would also be modeled. Would the resulting computer be stable? As it scales up, would it become existential and suicidal, or perhaps become an arrogant killer?
Several questions are raised about the aftermath of singularity. Can a downloaded personality replace the spirit? How does this equate to living forever? Singularity promises are similar to claiming that you can live forever by cloning yourself. One copy dies, but another digital copy survives. But it is clear that the copies are different entities.
And it is also clear that this is not true immortality. If we stretch singularity's approach to immortality a little further, we can argue that humans can achieve immortality through their work or art. And to this idea Woody Allen provided the best response: "I don't want to achieve immortality through my work. I want to achieve it by not dying."