The Philosophy Café

The Philosophy Café at Harvard Book Store is a monthly gathering meant for the informal, relaxed, philosophical discussion of topics of mutual interest to participants. No particular expertise is required to participate, only a desire to explore philosophy and its real-world applications.

The Philosophy Café is held on the third Wednesday of each month, from 7:30-9:30 pm, in the Used Book department on the lower level of Harvard Book Store.

Topic of Discussion for May:

In this cafe meeting we will explore the possibility of human-like artifical intelligence and examine at least three questions:

1. What are there fundamental limits to artificial intelligence? How do they compare to the limits of natural intelligence? Is it possible for people to construct a machine as intelligent or more intelligent than themselves?

2. Is it necessary or advantageous for such an intelligent machine to have such human mental qualities as awareness and emotion? Would it be possible in principle to build such a machine?

3. Can computers be imaginative and original or is this not possible in principle?

Suggested Readings

The readings below review the progress that has been made so far and contrast that with philosophical investigations into the nature and possibility of AI.

Turing Machines

In 1937, before any practical computers had been built, Alan Turing developed an elegant theory of computability and of a universal programmable computer, now called a Turing machine.

            see Wikipedia: http://en.wikipedia.org/wiki/Turing_machine

Turing also dealt with the problem of determining whether or not a program would always produce a result and halt. He proved that the problem was undecidable -- i.e. a Turing machine could not be programmed to determine if a  program would halt.

            see http://en.wikipedia.org/wiki/Halting_problem

Turing also considered building intelligent machines and proposed the famous Turing test as way of determining if a machine were intelligent. Here is link to his famous 1950 article, Computing Machinery and Intelligence.

            http://www.loebner.net/Prizef/TuringArticle.html

Computers and Originality

One critique of artificial intellegence is that computers are only able to do what they are programmed to. Does this preclude a computer from being creative or original? However there are a number of techniques to extend the utility of programs to novel situations not forseen by the program's developer.

            Genetic Algorithms: http://en.wikipedia.org/wiki/Genetic_algorithm

            Agent Systems: http://en.wikipedia.org/wiki/Multi-agent_system

Introspection and Generalization

A system with introspection is one that includes itself in its model of its world (problem space). The purpose for doing this is so the system can reason about its own reasoning to develop useful generalizations

            http://www2.imm.dtu.dk/~tb/defence.pdf

            http://www.iiia.csic.es/files/pdfs/1691.pdf

Importance of Context

Many problems cannot be solved by bottom-up techniques. For instance a simple isolated word spell checker cannot detect the misspellings that produce legal words, such as spelling from as form. A wider context has to be considered. The rise of the web has been very important in this regard, because it provides a very large and accessible data set of machine readable context. As a result, techniques for considering context have greatly improved in recent years. Web search engines, machine translation and IBM's Watson all owe a great deal to this development.

            Google's Computing Power Refines Translation Tool  

            http://www.nytimes.com/2010/03/09/technology/09translate.html

            Comparison of manual and machine translation          

            http://www.nytimes.com/interactive/2010/03/09/technology/20100309-translate.html

Critiques of Artificial Intelligence

            John Searle; Chinese Room Argument against the possibility of artifical intelligence

            http://plato.stanford.edu/entries/chinese-room/

            Hubert Dreyfus and his fundamental critiques of AI:       

            http://en.wikipedia.org/wiki/Hubert_Dreyfus

            J.N Hooker; A Postmodern Critique of Artificial Intelligence

            http://repository.cmu.edu/cgi/viewcontent.cgi?article=1209&context=tepper&sei-redir=1#search="artificial+intelligence+critique"

            Chalmers, French, Hofstadter; A Critique of Artificial Intelligence Methodology

            http://consc.net/papers/highlevel.pdf