Would you trust AI?
The Singularity (A.I.)
Since the first humanoid species walked the earth, mankind has influenced plenty of changes that have sped up the process of societal evolution. Humans have a far more superior brain than other creatures, granting us better cognitive ability, improved communication and social organization. Mankind is, therefore, able to think abstractly, work collectively and gather information used throughout the ages. These traits have enabled humans to modify the environment to suit daily living. Information technology is one discipline that has evolved greatly along with civilizations since the 20th century. Advances in information technology include: the proliferation of the internet, reduction in size of computing devices, Computer-Aided Manufacturing (CAD/CAM), Augmented Reality and Artificial Intelligence among others. This research focuses on the consideration of one controversial aspect in Artificial Intelligence – The Singularity.
The singularity is a hypothesis that has been put forward by various futurists and Artificial Intelligence enthusiasts. The hypothesis suggests the creation of an Artificial Superintelligence that will trigger uncontrolled technological growth, profoundly changing the human society (Kurzweil). While the concept of the singularity sounds like something out of a sci-fi movie, advances in Technology have led us closer to an age where machines are smarter than humans. A look at the advances in computing since the 1950s shows an exponential rise, denoting the possibility of an intelligence explosion. Two theories have been put forward to explain the exponential improvements in Information Technology: the Law of Accelerating Change and Moore’s Law. We explore these theories in relation to a rise in Superintelligence.
In the 1960s, Gordon Moore noticed that the number of transistors in a dense network doubled every two years (Moore’s Law). While this was just an observation, it has become an IT industry standard, with companies using the information for Research, Development and long-time product planning. Various authors and IT experts have further extrapolated and generalized this law, extending its application back to earlier integrated circuits, transistors and mechanical computers. For over half a century, Moore’s law has guided the evolution of electronic products, particularly computers, whose computing power has doubled every 18 months since then. The implications of Moore’s Law are apparent in modern electronics: improved memory capacity, better sensors, increased number and size of digital camera pixels and quality-adjusted prices of microprocessors (Bostrom). All these point to improved machine intelligence.
The law of accelerating changes marries the ideas in Moore’s law with that from the Law of Accelerating Returns. Ray Kurzweil postulated an exponential increase in the rate of change of evolutionary systems. An analysis of past trends in IT technology shows shortening gaps between advances. Whenever technology encounters an obstacle, newer advances help break the barrier. The law of accelerating returns can explain the sociobiological evolution that has occurred since the first human walked the earth. Since the dawn of evolution, creatures have developed faster and with more sophisticated traits geared toward efficiency (Kurzweil). Human progress has also been exponential, with the development of tools, language, philosophy, education and Information Technology to improve the quality of life. This pattern will result in unfathomable advances in the 21st century, possibly including the creation of the first fully functional machine intelligence.
Judging by the recent advances in modern computing, we see that the singularity is not only plausible, but also a necessary part of our sociobiological evolution. Three particular revolutions currently under research could propel us to runaway technological advancement: Genetics, Nanotechnology and Robotics (GNR). We are still in the early stages of genetic evolution. The study of human genes grants access to key information processes regarding life (Kurzweil). Using this information, we can reprogram our body to fight disease and the aging process. While studying genetics could enable the creation of near-immortal humans, this will be no match for the intelligent machines we could create once we fully understand the basic biology of life.
With nanotechnology, we could potentially reconstruct our physical bodies molecule by molecule. Nanotechnology involves engineering particles less than 100 nanometers in size, thus it finds application in molecular and atomic manipulation (Bridgstock, Butch and Forge). Therefore, nanotechnology could help us create bodies and brains with capabilities far beyond our chemical, physical and biological limitations. Robotics aim to create human-like machines with intelligence like our own. Robotics may be the biggest evolution towards the singularity, as sufficiently advanced A.I will be smart enough to predict and eradicate any obstacles in the path to superintelligence.
So, could a superintelligence unfold with these three evolutions? Combining these three disciplines, we could infinitely improve the capabilities of the human body and the machines we create. Genetics will help us understand the biology of the human brain and body. Therefore, genetics will enable us to achieve radical life extension by slowing down diseases and the aging process. Advances in biotechnology will also help realize gene-altering therapies, which could eradicate hereditary diseases and improve drug manufacturing & administration. Nanotechnology will help break the barriers to a genetically-induced biotechnological evolution. Assemblers constructed using nanoparticles could create self-replicating systems, which could be used in manufacturing almost anything on a molecular level (Clark and Eddy). Armed with the techniques of genetic modification and nano-level assembly, we could create robots with human-like bodies but with capabilities far beyond our own.
While we are at least a couple of decades away from achieving Artificial Superintelligence, current advances in automation, data science and biotechnology indicate that the singularity is not so far away. We discuss these disciplines in further detail:
Automation is a general term depicting processes that are performed without human intervention. Each waking day, more industries are accepting the use of programmed machines to achieve consistent and precise results. In industries, the spread of Computer Aided Manufacturing (CAM) has gradually reduced the workload on humans, instead using arithmetic control and algorithms to guide tools and equipment. Self-drive vehicles have also gained popularity over the past few years. Companies such as Tesla motors and Google have been working on cars that rely on GPS information, sensors and on-board computers to navigate roads and highways (Clark and Eddy). The Internet of Things (IoT) has also enabled the creation of smart systems via an internet connection. Homes, Power Grids, Retail shops and wearables use the internet to help ease daily living.
Data Science enables the creation of smart processes and systems. This discipline mainly identifies patterns within given sets of data. Human beings use data from the past to help guide decisions. In intelligent machines, we can use data patterns to create algorithms that help the system improve its efficiency (Machine Learning). Machine learning algorithms form the basis of most near-intelligent systems we have today (Kurzweil). Data science also helps improve access to information on the internet. Search engine algorithms such as the ones used by google utilize statistical methods to respond accurately to queries. Data science is, therefore, an integral part of any system deemed to be intelligent.
Ever since scientists mapped the human genome, we have made numerous advances in biotechnology. DNA sequencing has enabled us to identify the genetic makeup of viruses and other diseases, enabling the creation of effective and targeted remedies. Biotechnology has also enabled the formation of therapies that alter gene expression (the process by which genetic data is used to synthesize proteins). These therapies can therefore help eradicate deformations that are genetic in nature. Such therapies include: RNA interference, Gene chips, Cell therapy, and Somatic gene therapy (Bridgstock, Butch and Forge). These therapies could go a long way in improving human longevity through reversing degenerative disease, overcoming cancer and combatting heart disease. Bioengineering will, therefore, help us create smart systems with physical bodies that exist disease-free.
Just like any technological advancement, the rise of a singularity presents a set of benefits and challenges. This section explores the merits and demerits of developing a superintelligent AI.
One of the greatest advantages of a Superintelligent AI is its power consumption. Nanotechnology promises to improve the efficiency and capacity of computing devices. Devices built using nanoparticles are likely to obtain their energy from renewable sources such as solar, wind and geothermal technologies. Therefore, a superintelligence based on nanotechnology will reduce the strain on our energy budgets. Achieving the Singularity will also imply less risk of harm to human life. Not only shall these systems predict and eradicate life-threatening occurrences, they shall also replace mankind in energy-intensive, high-risk activities (Kurzweil). The Singularity also promises reduced research and manufacturing costs. A robot made to emulate a scientist/mathematician required nearly a third the cost of using human researchers.
A successful singularity will also offer the benefit of improved speed. Current desktop supercomputers can perform nearly 900 billion calculations per second. While this rate is desirable for desktop applications, nano-based computers offer a much higher potential, with the superintelligence promising near-infinite computational speed. The Superintelligence shall, therefore, solve complex human problems within a fraction of the time it takes researchers to find a breakthrough (Bostrom). The superintelligence will also use the large volumes of data publicly available to determine trends and establish solutions to mankind’s problems that remain unexplored.
The science community has aggressively pursued the creation of an Artificial Intelligence because of the benefits it promises. Critics are, however, quick to point out flaws that these enthusiasts often overlook. A superintelligence has the ability to self-replicate and build better systems within moments of deployment. Mankind could potentially be left at the mercy of these systems, probably making the superintelligence man’s last invention. Advances in computing and data science also reduce man’s learning capabilities (Carr). Since the proliferation of the internet and the invention of search engines, we can access all the information we need at the touch of a button. Scholars and literature lovers have admitted to losing their reading culture due to the ease of access. Nanotechnology and genetics could aid bioterrorism if the technology falls into the wrong hands.
Another demerit of the singularity involves its implications on life extension. Transhumanism and immortality will affect the quality of life and human psychology. We strive to achieve our personal ambitions knowing that we shall die (Clark and Eddy). With the singularity, however, this will not be possible since we have no deadline. One controversial issue with the singularity is that of job security. In many fields today, computers are already replacing human workers. Superintelligent machines could potentially take the place of entire work force or create lesser machines to do these jobs.
Even with these demerits, achieving a successful singularity is favorable and crucial to societal evolution. Judging by the evolution of economies, societies and technologies, the singularity is close. This implies a period of unprecedented economic and social growth. Improvements in the three technologies: genetics, nanotechnology and robotics, could further propel us toward a period free of disease, hunger, aging and even death. Thus, we should brace ourselves for technological singularity.
- Bostrom, Nick. Superintelligence: Paths, Dangers and Strategies. Oxford: Oxford University Press, 2014.
- Bridgstock, Martin, et al. Science, Technology and Society. Cambridge: Cambridge University Press, 1998. Document.
- Carr, Nicholas. The Atlantic. July 2008. 13 March 2018. <https://www.theatlantic.com/magazine/archive/2008/07/is-google-making-us-stupid/306868/>.
- Clark, Richard and R Eddy. Warnings. Harper Collins Publishers, n.d. Document.
- Kurzweil, Ray. The Singularity is Near. Amazon, 2005.
Gwendolym Dolske, Ph.D. KU Leuven
Science, Technology, and Society
Leave a Reply