ITGS Syllabus

Thursday, November 02, 2006

Topic 165

Ethical decisions regarding the use of robots in situations that might endanger human beings by Dwarkesh

Computers are the part of our culture. We have expectations that those computers think like machines. In addition, we are inclined to attribute intelligence to them. All these things were conceived in recent time. But what is decisive is the machine-machine interface. The Turing test concerns the man-machine interface, takes only 20 seconds, but that is not sufficient time for a genuine evaluation of machine. Machines with an experimenter can be trained.

A programmer can never anticipate what an interrogator may say or ask. The power of a computer can be only understood in the context of the social groups to which it is assigned, and human beings then compensate for the defects from artifacts. What can cause or teach and carry out an expert system for a social group? Nothing can be completely described; otherwise, machines could take over the competence-oriented jobs of workers? Well, this can apply, but this applies alone to behavior-specific competence. The only error is to believe that a tool is an actuator - i.e., an action.

There is a substantial difference between the way in which humans act and the way in which
machines copy a document. Intelligent machines are, however, the most useful and interesting tools. It is to be considered that robots can have the shape of humans and thus be able to simulate humans; this is hardly the case now, but it will be the case in the near future that their abilities will be similar to humans. They need the correct environmental model, thus they are not environmentally open (friendly) like humans; and they need a constant control and maintenance and they need stable environment conditions, in order to do the correct behavior for the day.

Probably, the instinct-secured behavior of some animals is more autonomous and more similar to humans than talented robots in the foreseeable future. So far automats with only very limited functionality can be realized, because they lie within a range of pre-rational intelligence, within which certain human competence is to be simulated. The robots of future will be copied still more from their models, humans or animals. And in robots the artificial and biological elements will unite, since computers, especially networked computers, are disembodied Artificial Intelligence.

The realization of this intelligence in a body-free Cyber world (a sort of disembodied world) is a possibility on one side, and the realization in robots is on the other. But we will still have to wait and see which way will be best for AI development.

For artificial intelligence to become true intelligence, it must become artificial soul and not
be limited to a pure mind or pure cognition. In addition, it's necessary to understand that
feeling and motivation actually contribute to the increase of intelligence to a considerable
degree and that they are possible in the computer.

The free Will is the released Will, or the released motive. Man will build artificial souls, because we are making an attempt to copy the psychological processes to give us much greater understanding. Human beings should know substantially better, and that is so far the case whether, however, a human soul could be designed independently of a human body in a form other than an extremely rough model of a soul, I think is very doubtful. Besides, it would have to be clarified as to which psychological characteristics are to be given to such a soul.

The desire to imitate humans can represent a specific perspective for robotics. It is objected that the communicative competence of humans is not in principle technically replaceable. Technology proceeds right from the outset, where it substitutes for human activities, procedurally differently than nature.

Cognitive robotics is for engineers, at the beginning of technical conversion of a new body-soul theory -- functionalism. But it ignores the role of humans, who can not be simulated. In order to be able to take responsibility, technical systems must be controllable by humans. By a complementary system, organization work quality -- in the double sense of competence promoting tasks and effective achievement -- contribution becomes possible. Occasionally, one must consult the collectiveness of beneficiaries, in order to compensate the damage to those the technology hurt by malfunctioning. This refers both to producer and consumer.

Computers are today, in symbolic algebra, better than humans. Computers with the correct software actually think about facts, make decisions and have goals. The task is to build robots that act as if they seem to have fear, or simulate the fear. We draw a limited comparison with our feelings: in the animal realm we are still safe in our place, and so we confess some feelings to animals. It is similar for machines, and the machine operates in the same way as humans. Artificial Life (AL) and the artificial evolution create whole new thought categories.

Models of neural networks still need to be developed, and the question of consciousness is a very difficult question indeed. The idea of robot as our slaves is a beginning of morally justified slavery. And so, if robots are to have the status of humans, then this form of slavery must also be eliminated. In 20 years our personal computers will be 1000 times more efficient than today and will then exceed our intellectual capacity. Perhaps we can build a computer that is more intelligent than we are, and perhaps not.


Post a Comment

<< Home