Stage-1: 50-100 Years
Homo Computerus is the name we have given to the evolutionary stage that follows Homo Sapien. This name is fictitious and, to some extent, slightly misleading in the sense that the beginnings of this evolutionary stage pre-dates computers. The other aspect, which may be contentious in terms of classical evolution, is that Homo Computerus is essentially genetically identical to Homo Sapien. While there may have been virtually no major change in human DNA for several hundred thousand years, clearly an incredible evolution has taken place in terms of human civilisation over this same period. A historical timeline lists some of the major achievements of humanity starting with primitive tools, the ability to make fire, build canoes and construct looms for weaving. These achievements were the building blocks of early civilisations, which started to emerge some 5-10 thousand years ago. By the 16th and 17th centuries, empires spanned the globe and science had proved that the Earth was a planet circling the sun. Subsequently, the 18th and 19th centuries saw the consolidation of the industrial revolution and the ‘engineering’ of a new type of civilisation. Finally, from the current perspective, the evolution of civilisation has now culminated in the ‘Information Age’ due to the development of computers and telecommunication.
So what was the catalyst for all this phenomenal change?
The catalyst that caused the evolutionary step towards Homo Computerus was the ability to better communicate information, but more precisely knowledge. Of course, we must give natural selection the credit for the increased brain capacity and vocal ability in humans, as it was these capabilities that were to start the dramatic increase in the amount of knowledge that could be passed on from one generation to the next. Eventually, the development of the written word and the invention of the printing press would lead to more fundamental changes in the ability to communicate information more effectively. The evolution of human civilisation is directly reflected in the ability of each generation to assimilate and acquire knowledge. However, it is argued that the emergence of computer technology in the 20th century was the major catalyst, which signalled the arrival of Homo Computerus, hence the name. Although Homo Computerus is not a new species within any accepted evolutionary classification, it is being suggested that unlike Homo Sapien, Homo Computerus would become extinct without computers. In a sense, Homo Computerus highlights the fact that a transition from natural to artificial evolution has already started to take place. In practice, this transition is not yet complete, as its definition will continue to grow and come to encompass the development of weak AI systems brought about by the maturing of several new fields of science and technology.
How might the evolution of Homo Computerus be significant?
Although the hybrid combination of human and computer intelligence is not yet physically integrated within Homo Computerus, its intellectual capacity is already orders of magnitude greater than Homo Sapien in terms of the breadth and depth of knowledge. The ability to maintain, distribute and exploit this knowledge is now one of the key factors to further progress. However, there is considerable potential for weak AI systems to develop over the next 50 years. One of the most significant aspects of hybrid AI will be the intelligent management of information by weak AI, which will effectively augment the creativity and imagination of human intelligence.
We are still discussing the first stage of evolution toward AI, but there is a need to step back and ask possibly a more a fundamental question relating to intelligence.
What makes a genius?
On one level, we could assume a genius is more intelligent because their brain is simply larger, although physical examination does not really support this notion. In truth, like so many questions concerning human intelligence we simply, as of yet, do not know the answer. However, AI research is beginning to understand that ‘common sense’ requires an ability to make associations between explicit and implicit information to derive a new solution. In these terms, a genius might just be a person who has acquired the right combination of explicit and implicit knowledge plus ‘sufficient intelligence’ to make a new association. As such, Homo Computerus may appear as a genius to Homo Sapien, not because they are intrinsically more intelligent, but because they possess the technical ability to store vast amounts of explicit and implicit knowledge and have the capability to make millions of automated associations per second. In this respect, the combination of human ingenuity and weak AI could create 100 geniuses every year, instead of 1 genius every 100 years. On this assumption, let us consider how science is developing its ideas about information:
- Memetics: Theory for understanding the spreading of information
- Cybernetics: Control of information in organised systems
In biology, DNA information is encoded into ‘genes’ and is physically transferred to future generations by way of its direct descendents. In higher life forms, information can also be communicated through actions or words. A ‘meme’ is the name given to a unit of transferable information and the collective process is called ‘memetics’. Richard Dawkins first introduced the concept of memes and memetics in 1976, when noting that cultures can evolve in much the same way as organisms, i.e. by passing information from one generation to the next. He also noted, that some ideas can increase or reduce the probability of survival of an individual, which in so doing, actively filters some of the ideas that get passed on to future generations. However, while the process of replication of memes has some parallels to genes, the transfer of information is much faster and not restricted to direct descendents. In fact, this process of information transfer is more analogous to a viral infection. At this stage, memetics is not a science, but more of an idea that could act as an important catalyst for science that crosses the traditional subject boundaries. This is particularly important to AI in general, as it is believed that many significant breakthroughs will require knowledge that cuts across today’s vertical segregation of the sciences. For example, a biologist or a computer scientist could be equally interested in interpreting the following characteristics of information transfer, albeit from different perspectives:
- Accuracy: If errors occur in transfer, then information may either become damaged or dangerous.
- Distribution: The spread of information depends on speed and size of the distribution mechanisms available.
- Longevity: Can be a function of accuracy and distribution plus the durability of the containment vessel sourcing the information.
In the past, the main institutions of society supported great libraries that represented the sum of human knowledge. The collation and preservation of this information was a slow and expensive process, restricted to the privileged few. Today, the Internet in the form of the Web starting to emerge and presenting information to new generations in revolutionary new formats. However, like much in today’s society, the Web is probably being over-hyped, as it is only predicated on very weak AI principles that are still in infancy. Some critics even complain that the Web is only delivering junk mail at the speed of light, and while there is some truth in this position, it ignores the latent potential. Today, the issue with information is one of both quantity and quality; as such, there is a growing requirement for intelligent control. Weak AI, in combination with cybernetic principles, could be a major catalyst that effectively increases the intelligence on planet Earth. If this is the case, then subsequent generations of the Internet could allow huge amounts of information to be semantically indexed and the search for new knowledge associations to be automated. These systems would not be intelligent in the true sense of the word, but serve to augment the efficiency of intelligence in a hybrid AI world. Hence the potential importance of cybernetics as a science that looks to control and enhance information.
Cybernetics was initially developed to help optimise the transfer of information through telecommunication channels via the use of feedback concepts derived from engineering control systems. However, cybernetics has grown to encompass many of the leading-edge ideas in computer intelligence:
- Machine learning
- Man-machine interfaces
- Autonomous agents
- Neural networks
While the very term ‘cybernetics’ may conjure up futurist visions of intelligent computer technology, its roots are still linked to systems that require the control and distribution of information. However, it is clear that weak AI systems could have a profound impact on the design of information control systems in the next 50 years:
- Cybernetics is predicting that expert systems, machine learning,
and neural networks that contain some cognitive processes will automate
- In addition, new computer-based applications will revolutionise the organisation and representation of information in the format of useful knowledge. These tools will also support the building of information models, such as simulations that also use virtual reality, hypertext, multimedia, databases and semantic information retrieval.
These features of weak AI systems will allow knowledge to be intelligently structured to the requirements and preference of an individual. At the same time, the automation of communication and control will lead to the eventual maturing of the Web as a network of weak AI systems from which the concept of ‘cyberspace’ could truly start to emerge. Of course, all these predictions are predicated on a level of computer processing that does not exist today. However, there is good technical evidence that computer processors will continue to develop until about 2020. If so, processors will be a 1000 times faster than today. It is also reasonable to assume that other aspects of computing; such as memory capacity and communication speed would also improve over this period. Beyond 2020, further progress may possibly depend on the development of radically new architectures. Although, at this time, it would only be speculation, as to which of the possibilities might win the day, it is not necessarily wishful thinking to assume that one will emerge. It is also reasonable to assume that new software architectures will develop in-line with the capability of the underlying hardware processors, especially if accompanied by a better understanding of the symbolic and pattern recognition processes within the brain. Therefore, based on the relatively high probability that significant improvements in both hardware and software will continue, weak AI systems will also continue to improve year-on-year.
Although the doubling of computer processors every 18 months is a well talked about factor, potentially it is the growth in the breadth and depth of knowledge spawned by computers that is the more significant factor. Even comparatively weak AI systems could revolutionise the processing of information and act as a feedback mechanism that leads to a further acceleration of progress. Therefore, the timeline proposed for the AI paradigm could be conservative from a technical perspective, but the braking effect that some social institutions, not to mention the economy, may need to be taken into consideration. However it seems almost inevitable that humanity will continue to disappear down the road towards hybrid AI. For now, like the cowboy movies of old, we will assume the road leads towards the sunset and a better life.