BOOK TWO: NEW HUMAN BEINGS, 2020 - 2060

Chapter Eleven: Mechanisms Of Human Evolution In The 21st Century

V. Robotics

Running alongside the development of bionic improvements to the human body and mind will be the development of robots. Robots need to be considered in tracing the future course of human evolution because they will come to be viewed in many respects as supplementary to our existing bodies and minds.

So much attention has been paid to the likely nature of robots in literature (science fiction) and movies, that they have become a part of folk psychology even before they exist. Everyone thinks they know what robots will look like, what they will be able to do, their probable use by governments to oppress people in general or just to control and 'take out' criminals. Many people probably even know the laws of robotics as laid down by science fiction writer Isaac Asimov in 1950.

It is normally assumed that robots will be given physical skills and mentalities comparable to those of humans, but this is far too simplistic an assumption, especially given the hodge-podge of behavioural traits humans have developed, ranging from murderous aggressiveness to altruism.

At the physical level, there may well be considerable similarities, at least at first. Although non-humanoid intelligent robots already exist in environments such as automotive manufacturing, it is likely that most robotic development will parallel human forms in its early stages, since robots are most often conceived as machines which will extend the reach of existing humans. That suggests a lot of commonality between the design and control of robotic elements such as fingers, joints, limbs, and of course their sense organs: sight, hearing, touch and taste. It may be expected therefore that where parallels exist, robotic devices will largely mirror bionic ones, or vice versa. Robotic joints will look like human ones, especially bionic human joints.

Designing the cognitive equipment of humanoid robots will pose greater challenges. Although much progress has been made already in terms of understanding what one might call the flow charts of human decision-making – the interaction of sensory input with data-bases held in the brain, the carrying out of a decision process, and the implementation of the decisions made through efferent nerves (or through speech – an alternative way of giving commands to an external unit), it would not be right to say that the 'wiring diagrams' of the processes involved in the brain have been deciphered to a point at which they could be copied for robotic purposes.

Nor can it be assumed that they ought to be copied, even if they were known. The human brain is a mish-mash of component sections, many of which came into existence long before primates existed, and some of which have been re-used or developed for newer purposes. Although nature does reach the most efficient solution to cognitive problems, it only does that within the existing structure of the brain. Even if a robotic control function is exactly equivalent to a human control function, it doesn't necessarily follow that it should slavishly follow the neural structure of the equivalent function in a human brain. In addition, robotic control functions will often deliberately not follow human control mechanisms.

Deciding whether to map the groupish skills and attributes of humans onto robots will be a particularly ticklish process! For example, the possibility of damage to con-specifics is taken into account in human decision-making, whereas robots (according to Asimov's laws) should put the integrity of humans at the forefront of their decision-making. On the other hand, it is easily imaginable that a robotic designer might want to build reciprocal altruism into a robot programmed to deal with children. Even then, of course, filleting out reciprocal altruism in a human brain may or may not ever be possible in neural terms (too early to say). It may be that, once the relationship of such a trait as reciprocal altruism has been adequately mapped in terms of its functional connections to other cognitive attributes, it will be necessary only to duplicate those connections in terms of robotic electronics. On the other hand, it may be simpler to define a trait such as reciprocal altruism in behavioural rather than neural terms, and design it into the robot in that way. For some traits, there may turn out to be no distinction between behavioural and neural mappings, which would be helpful!

Those who fear that robots might 'gang up against us' are probably wide of the mark; but one can see that designers would be tempted to arrange that humanoid robots would view their own con-specifics as 'machines' rather than having humanoid characteristics, and to avoid building affiliative drives into robots. Such issues are explored further in Chapter 12.

By, say, 2030, the functional cognitive structure of the human brain will be well understood, although some 'wiring diagrams' will not yet be mapped, not least because they are dynamic. The issue of the extent to which words are stored in terms of related images or non-linguistic patterns will have been resolved, and appropriate results will have followed in terms of robotic and bionic cognitive devices, and the control of them. The main lines of the cognitive structure of 'human-friendly' robots (ie robots which need to communicate in more than a superficial way with humans) will have been laid out.

During the period from 2030 to 2050, it is to be expected that humans will become able to communicate with quasi-human robotic intelligences using wireless or magnetic technologies, or just using a cable link, by-passing their normal sensory channels. It will probably be possible for a human to 'inhabit' a robot's psyche, using its consciousness, its cognitive tools and its sensory equipment as if they were her own. The opportunities that this will offer, particularly for expanding human groupish behaviour, are explored in later chapters.

Alongside humanoid robots, other types of robot will be attracting increasing attention during the early decades of the 21st century. The automated roving scouts of search engines on the Internet are already called 'bots'; of course they are purely disembodied, electronic robots. Some of the uses for which robots with specialized and distinctly non-human cognitive structures (not to mention physical or electronic forms) will be developed might include:

Of course this is not a complete list. Many of these types of robots have already begun to exist or be developed. It may well be that some of these robots will be given quasi-human appearance, although they are probably going to be better at what they do if they are equipped with specialized cognitive structures that are designed from the ground up rather than being based on human originals. Humans did not evolve to carry out any of the tasks listed above, even if they have become adequately good at some of them. It may also be that robot designers will choose or be required to include robot-human communication interfaces to allow humans to 'interfere' in the robot's own decision processes if necessary. Robots, however specialized, are also going to need to be aware of humans as part of their environment, and treat them appropriately; but that doesn't mean they have to learn to smile, unless we want them to!

By 2050, all of this will have happened, although neither robots nor bionic super-people will be run of the mill. There is more, however. To get a full picture of how people and their robotic inventions may look by 2050, it is necessary to take into account genetic changes, a subject which is approached in the next section.