Roboethics - Ethical Engineering for International Development and Environmental Sustainability - page 73

differentiates human intelligence from machine intelligence. We already have
computers with computing power many times greater than that of a human, but they
are clearly not intelligent. Relevant factors include creativity, fl exibility, independent
thinking and probably also unpredictability. If meaningful technological representa-
tions of these qualities are feasible, which is still an answered question, their com-
bination with high computing power could lead to a very formidable and possibly
highly dangerous intellect. Artifi cial intelligence approaches used to investigate this
problem include the following: (1) perception and analysis of the environment,
(2) natural language processing, (3) human interaction, (4) cognitive systems,
(5) machine learning and behaviour and (6) neural networks.
An important feature of intelligent robots is their ability to learn, for instance, the
characteristics of the surrounding physical environment and the animals and people
that inhabit it. In addition to learning about their environment, robots should be able
to learn about their own behaviour, through a self-refl ective process. This will
include learning from experience, replicating the natural processes of the evolution
of intelligence in humans and animals, including synthesis procedures, trial and
error and learning from experience. Intelligent robots would require the ability to
learn about ethical issues and the consequences of their decisions and actions.
Figure 3.3 shows one possible development path for robotics. This involves
changes from unintelligent industrial robots via intelligent industrial robots to
intelligent mobile, including humanoid, robots to the third generation of advanced
intelligent robots which are able to interact and work together with people.
This raises ethical issues related to both the robot's actions and the desirability of
such developments and their impacts on human societies. It is these latter impacts
which are the most important and the most worrying, as well as very diffi cult to
foresee. While it is likely that there will be threats and disadvantages, current
predictions may be incorrect and it is also diffi cult to imagine what the real benefi ts,

Fig. 3.3 Development trends in robotics (Kopacek 2005 )