Artificial Intelligence (AI)

In 1637 Rene Descartes, the French mathematician and philosopher, predicted that it would never be possible to make a machine that thinks as humans do. That was a rather astonishing observation considering that the concept of the analytical machine was devised by Charles Babbage only two hundred years later. Babbage never completed his analytical engine but his theories laid the early foundation for artificial intelligence.

The father of Artificial Intelligence was British mathematician Alan Mathison Turing. In 1950 he declared that in the future there would be a machine that would duplicate human intelligence. He devised a specialised test, known as the "Turing test", to be used to prove artificial intelligence. In the test, a human and a computer hidden from view would be asked random identical questions. If the computer was successful, the questioner would be unable to distinguish the machine from the human.

In 1947 Turing argued that the brain could itself be regarded as a computer. Working on his Automatic Computer Engine, he declared that he was more interested in producing models of the action of the brain than in the practical applications of computers.

AI laboratories
The first conference on artificial intelligence was held at Dartmouth College, New Hampshire in 1956. It led to the establishment of the AI laboratories at MIT (Massachusetts Institute of Technology) by Marvin Minsky and John McCarthy (who invented the AI computer language called Lisp) and Stanford University by Edward Feigenbaum and Joshua Lederberg. Herbert Simon and Allen Newell of the Rand Corporation ran tests that showed the one and zeros in computer language could be used not only to represent numbers but also symbols. Between 1958 and 1960 psychologist Frank Rosenblatt of Cornell University modelled the Perceptron computer after the human brain. He "trained" it to recognise the alphabet. The chase was on to develop "neuron networks" of computer processors.

The human brain consists of more than 100 billion nerve cells (neurons) through which the brain's commands are sent in the form of electric pulses. It can process many operations at the same time (such as thinking, talking and walking at the same time). This is called parallel processing. Computers follow sets of logic steps, procedures called algorithms. Fast computers perform roughly 10 billion calculations per second. Supercomputers use multiple processors to follow several algorithms simultaneously.

Back to the power of reasoning

When IBMs Deep Blue defeated world chess champion Gary Kasparov in 1997 it was a boost for AI developers. Today, a host of "smart devices" can recognise postal codes, patterns, symbols, handwriting, voices, etc. But no computer has yet mastered "plain, common sense." Computers, it seems, can talk to each other but not to humans.

If the computer is to think like humans then its brains should be developed to be like that of a human. So, instead of using digital processors, scientists have developed silicon chips that work in analogue mode, the way a human brain cell works. A computer at the Argonne National Laboratory in Illinois used this mode of operation to process highly abstract problems, crudely approximating human reasoning.

The androids

The idea of personal assistant robots are not too far off, perhaps. But how will these human-like robots, called androids, behave and how will they be governed? Won't they "take over the world?" If the robot laws of Isaac Asimov is followed, we'll be safe.
1. Asimov's first law is that robots may not harm humans either through action or inaction.
2. They must obey humans except when the commands conflict with the first law.
3. Androids must protect themselves except, again, when this comes into conflict with the first law.