Artificial Intelligence - Final Report (Section 11)

Artificial Intelligence

Photo Credit: GeekWire

"Every aspect of learning or any other feature of intelligence can be so precisely described that a machine can be made to simulate it."                                              - The Dartmouth Proposal

     This is a line from the famous Dartmouth Proposal, which is credited with introducing the term artificial intelligence.  As this statement suggests, many aspects of human intelligence can be simulated by machines.  This is evident in the AI that can beat master chess players, hold convincing human-like conversations, and perform many actions a human could perform, but machines have yet to simulate the more abstract qualities of human intelligence like free-will, thought, and consciousness.  This is a possibility that is the foundation of various movies like her, Blade Runner, and I, Robot and has ignited much interest in the field of AI.  This possibility has also drawn much philosophical interest and arguments, looking to truly figure out whether this possibility could become a reality.  Many of these arguments focus on whether machines can display general intelligence, or the ability to solve any problems humans can also solve.  This report will look into some of these philosophical arguments.

Visual Interpretation of the Turing test

Alan Turing and the Turing Test

Alan Turing was an English mathematician, computer scientist, and creator of the Turing test.  The Turing test is a test of a machine's ability exhibit intelligent behavior equivalent to human intelligence.  Turing does this by defining intelligence with a simple conversation.  A modern version of the Turing test would take place in a online chat room where one participant is a human, one participant is a human interrogator, and one participant is a computer program.  The human interrogator would ask questions to the machine and human participants and then make a decision on who is the man or machine.   If the interrogator cannot consistently tell which participate is the computer program, the computer program wins.  Here is a video of an AI that could pass the Turing Test:

If one were to have a conversation with someone they believed to be another person, it would be a polite convention to say that they were intelligent or believe they were talking to another intelligent being.  Turing extends this polite convention to machines by saying that if a machine can answer questions, using the same words a human would, then the machine would be considered as intelligent as a human. 

Searle's Chinese Room

The Chinese Room Photo Credit: Neurologicablog

     John Searle argues against the Turing test by proposing the Chinese Room argument. In his paper "Minds, Brains, and Programs", Searle asks us to consider the Chinese Room thought experiment: suppose there is a computer program that passes the Turing test and demonstrates general intelligent action. The program can also converse in fluent Chinese. Suppose the program instructions are written on cards and then give them to a person who does not speak Chinese. Lock the person in a room and have him follow the program instructions. He will copy out Chinese characters and pass them out of the room through a slot. From the outside, it will appear that the Chinese room contains a fully intelligent person who speaks Chinese.

     Though, does anyone or anything in the room truly understand Chinese? Does anything in that room have the mental state of understanding, or which has conscious awareness of what is being discussed in Chinese? Searle says if you look at all the pieces of the room, there is not an being with understanding of Chinese in the room since the man does not speak Chinese and neither do the instructions. Searle then concludes that the Chinese Room, or any other physical symbol system, cannot have a mind as stated by a quote in Searle's "derivation by axioms":

"(A1) Programs are formal (syntactic).(A2) Minds have mental contents (semantics). {A3) Syntax by itself is neither constitutive of nor sufficient for semantics."

                                                                                                                   - John Searle

     Searle goes on to argue that actual consciousness and understanding require "actual physical-chemical properties of actual human brains." He argues certain chemical properties in brains and neurons give rise to minds: as Searle once said "brains cause minds."

Symbol Processing

A physical symbol system takes symbols, combining them into expressions and using processes to produce new expressions.  These symbol systems can be seen in formal logic, algebra, and computers.  This idea was proposed by Allen Newell and Herbert A. Simon. They wrote:

"A physical symbol system has the necessary and sufficient means for general intelligent action."

— Allen Newell and Herbert A. Simon

This claim implies both that human thinking is a kind of symbol manipulation and that machines can be intelligent.

These ideas can be found in the ideas of many older philosophers like Hobbes - who claimed reasoning was "nothing more than reckoning", Leibniz - who attempted to create a logical calculus of all human ideas, Hume - who thought perception could be reduced to "atomic impressions", and Kant - who analyzed all experience as controlled by formal rules.

Implicit Skill and Neural Nets

     Hubert Dreyfus, an American philosopher and professor, argues that human intelligence and expertise depends primarily on implicit skill rather than explicit symbolic manipulation, and argues that these skills would never be captured in formal rules.  This is due to his belief that classical AI will never become a match for human intelligence due to classical AI's belief that human intelligence works off a set of formalized ideas like classical AI.  Dreyfus argues that the human mind functions intuitively and not formally.  These arguments have led to many artificial intelligence researchers away from using formal logic.

Though with the evolution of Computational intelligence paradigms like neural nets, evolutionary algorithms, and etc., new artificial intelligence algorithms that mimic unconscious thinking have grown. This has led to Dreyfus himself agreeing that these sub-symbolic methods can mimic the kind of "tendencies" and "attitudes" that he believes are essential for intelligence.

As seen from these arguments, philosophy has much to say in the field of artificial intelligence.  While these machines may or may not match the intelligence of humans in the future, these arguments and ideas have made researchers question their methods and evaluate the best ways to approach creating truly intelligent machines.  Hopefully, the fields of philosophy and artificial intelligence can continue to work together to create better intelligent machines in the future whether they have general intelligence or not.

Quiz Questions

1. Who created the Turning Test?
2. What proposal coined the term Artificial Intelligence?

Discussion Questions

1.  Do you believe an artificial being could become "human"?
2.  If you broke human thought down, do you think it would be just chemical and        electrical signals or is there something more to us?

Final Report Comments

https://cophilosophy.blogspot.com/2019/12/justin-hightower-section-12-final-draft.html#comment-form

https://cophilosophy.blogspot.com/2019/12/the-fear-of-death-travis-moreland.html?showComment=1575765828724#c7677204741785643142

Sources

Turning, Alan, The British Library, https://www.bl.uk/people/alan-turing/.

Oppy, Graham and Dowe, David,  Stanford Encyclopedia of Philosophy,  https://plato.stanford.edu/entries/turing-test/.

Hauser, Larry, Internet Encyclopedia of Philosophy, https://www.iep.utm.edu/chineser/#H1.

Kenaw, Setargew, Hubert L. Dreyfus’s Critique of Classical AI and its Rationalist Assumptions, https://link.springer.com/article/10.1007/s11023-008-9093-7.

Dreyfus, Hubert, 1992, What Computers Still Can't Do, New York: MIT Press.