Artificial Intelligence

     My Home Page

An extract from and essay written in september 2001

Artificial Intelligence (AI)

For centuries some of the great minds in philosophy, psychology and science have wondered how the set of processes we call the mind emerge from the activity of the organ we call the brain. Revealing the biological basis of the conscious mind is essential if we are to create AI. So are we close to these revelations? For some people conducting an investigation with the very instrument being investigated is a conflict between the observer and the observed. However, for others there is no problem in science that is insurmountable especially with relentless and exponential increase in new knowledge and ingenious new theories and techniques.

Economic forces have a great bearing on the development and speed of progress of AI. Much more investment and time is required to learn about our cognitive abilities and how we can transfer these to AI. We need to see the public on board. This will only happen if application become commercially successful and businesses benefit, and see the potential of further benefits from the applications of AI.

We have learned so much about the brain, especially in recent times. But we also must remember that the current description of neurobiological phenomena is incomplete and much more research is required in this area. But even in the nineteenth century Paul Broca and Carl Wernicke were able to show how different regions of the brain were involved in processing different aspects of language. Today, using brain imaging techniques such as PET (positron emission tomography) or fMR (functional magnetic resonance) we can directly record the activity of a single neuron or groups of neurons and relate that activity with aspects of a specific mental state such as perception of the colour red or of a curved line. We can reveal how different brain regions are engaged by certain mental efforts such as learning to identify a face. We can even determine how molecules in microscopic neuron circuits participate in mental tasks and identify the genes necessary for the production and deployment of these molecules. And we can see how newly learned facts are consolidated in long-term memory by strengthening or weakening the synapses between neurons.

These non-invasive brain imaging techniques will eventually be accompanied with invasive techniques. We will reverse-engineer the human brain. Already people are leaving their brains, as well as their bodies, to science. Our examination of neuron connections, and concentrations of neurotransmitters at the synapses will reveal vital information for understanding the cognitive processes. And in the future we will develop microscopic robots (nanobots) and send them into our bloodstream's to explore our inner circuitry. These nanobots will not be limited to passive roles such as monitoring. They will be built to communicate directly with the neuronal circuits in our brains, enhancing or extending our mental capabilities. Connecting neurons to silicon is only in its infancy, but we have already connected the two and caused the neuron to fire when instructed by the computer processor.

Also, we are tackling the problems of AI from many angles. Much research has been done on neural nets, an emulation of the computing structure of the neurons in the human brain as well as on genetic, or evolutionary algorithms which are based on allowing intelligent solutions to develop gradually in a simulated process of evolution. So perhaps we are not far off understanding cognitive processes and the traditional dualistic separations of brain and mind will fade. If we do succeed in unravelling the mysteries of the brain then there is no reason to think that we can't imitate the workings in artificial systems. I think it was not that the pioneers were wrong in their predictions but just a little too premature.

Can human-level intelligence be achieved by writing large numbers of programs and assembling massive databases of knowledge, facts and rules and running the programs on super-fast computers? Many AI researchers think not and believe that new fundamental ideas are required and that it is difficult to predict when, or if, human level intelligence will be achieved. Despite phenomenal progress in recent years, no computer yet devised even approximates in its capacity the powers of the human mind. Perhaps the computers of today are fast enough, and knowing how to program the computer correctly is the key.

The philosopher John Searle says the idea of a non-biological machine being intelligent is incoherent. The philosopher Hubert Dreyfus says that AI is impossible. The computer scientist Joseph Weizenbaum says the idea of AI is obscene, anti-human and immoral. Clearly, there is much disagreement amongst the great minds in the world today but I tend to agree with Ray Kurzeil's approach, with regard to the future. Artificial Intelligence is inevitable. The abilities of the computer and the species that invented it will grow even closer. While today we do not understand the basis of human intelligence and have therefore failed to create AI, we will understand it, and in the near rather than the distant future. Machine intelligence is here in infant form today and now it is learning to walk, talk and see. It will have the perceptual skills, inference and linguistic abilities that we possess as well as having unlimited memory, and excelling in attentional and reasoning abilities which we sometimes have difficulties with.

Will AI be conscious? Will it have the ability to have subjective experience, have free will and be self aware? AI will appear to be conscious but this will be an ongoing debate just as we debate today whether non-human animals experience consciousness. Only humans, chimpanzees and orangutans have passed what is known as the mirror test (though it should be mentioned that this is still being debated). And even children fail the mirror test (infant amnesia) until they reach about two years of age, when their pre-frontal cortex begins to mature in structure and function. Passing the mirror test means that some animals are aware of themselves and can infer mental states of others. They can sympathise, empathise and attribute intent and emotion in others - abilities that were considered to be the exclusive domain of humans. As the solitary lifestyles of early primates gave way to the existence of large groups, the resulting complex social interactions enabled these abilities to evolve.

Human intelligence was created from 'nothing' over vast periods of time. If it was possible to create it naturally, by a process of evolution, then it should be possible to create it artificially - when we can speed up evolution a million fold. We are slowly but surely unravelling the secrets of the universe. If it turns out that we cannot create intelligence artificially then I think this is a more profound discovery than if we succeed. If, or when, we do create AI it will mean a profound revolution in human self-understanding.

And to quote Richard Dawkins from 'The Blind Watchmaker', "There is a popular cliché which says that you cannot get out of computers any more than you have put in, that computers can only do exactly what you tell them to, and that therefore computers are never creative. This cliché is true only in a crashingly trivial sense, the same sense in which Shakespeare never wrote anything except what his first schoolteacher taught him to write - words".

back to top of page  Back to top of page