Churchland is exploring the past and current views of AI in order to further explore the future of AI. They believe that AI cannot currently be conscious, but that future mapping and reverse engineering may produce an AI equivalent of the human brain.
The first argument was whether or not a machine could pass the Turing test. It was argued that if a machine had enough memory and was programmed correctly it could pass the Turing test. Results showed that a machine could pass complex tests that were presented. However, the argument was made that these tests were still simply programs. The machine still had to receive relative input to produce a given output. It could not create an output based on intuitive syntax.
A second argument was presented that although a machine may only be manipulating symbols in a rule-based environment, it may still be “thinking” only on a simpler level than what we might consider intelligent. They argue that simple thinking is still thinking.
Churchland then explores intelligence through the structure of the brain. They attempt to explain the processes of the brain through reverse-engineering. Their biggest argument is that the brains neural network functions on a parallel level, through millions of pathways simultaneously while current machines function on a serial level. And that this complex neural network is beyond anything that could be constructed artificially. This network is so complex and so fast that it functions almost instantaneously. They also argue that to produce an intelligent machine it would have to be able to function as a complete human brain due to the way human intelligence is created. There is simply no other way to use this hardware/software combination.
In the end, Churchland rejects the Turing test as a “sufficient condition for conscious intelligence.” They base this conclusion...