05-turing-test.pdf

Transcript

1. Can machines really think?
   The Turing Test and
   the Chinese Room
   George Matthews
   CC 2017
   

   View Slide

2. From Functionalism to AI
   

   View Slide

3. From Functionalism to AI
   

   View Slide

4. From Functionalism to AI
   If minds are really just programs running
   in our human hardware, then it should be
   possible to build a thinking machine, an
   Artificial Intelligence, shouldn’t it?
   

   View Slide

5. Possibilities
   

   View Slide

6. Possibilities
   

   View Slide

7. Possibilities
   Logical: there is no contradiction in the idea of a
   thinking machine.
   

   View Slide

8. Possibilities
   Logical: there is no contradiction in the idea of a
   thinking machine.
   Physical: thinking machines do not violate the laws
   of physics.
   

   View Slide

9. Possibilities
   Logical: there is no contradiction in the idea of a
   thinking machine.
   Physical: thinking machines do not violate the laws
   of physics.
   Technological: we know how to actually build a
   thinking machine.
   

   View Slide

10. Possibilities
    Logical: there is no contradiction in the idea of a
    thinking machine.
    Physical: thinking machines do not violate the laws
    of physics.
    Technological: we know how to actually build a
    thinking machine.
    Social/Moral: there is no strong reason we should
    not build a thinking machine.
    

    View Slide

11. The Turing Test
    

    View Slide

12. The Turing Test
    How can we tell whether a machine is intelligent?
    

    View Slide

13. The Turing Test
    How can we tell whether a machine is intelligent?
    ! Don’t we have to define intelligence first?
    

    View Slide

14. The Turing Test
    How can we tell whether a machine is intelligent?
    ! Don’t we have to define intelligence first?
    ! Don’t we have to “reverse engineer” ourselves to
    build an intelligent machine?
    

    View Slide

15. The Turing Test
    How can we tell whether a machine is intelligent?
    ! Don’t we have to define intelligence first?
    ! Don’t we have to “reverse engineer” ourselves to
    build an intelligent machine?
    Turing’s answer: we do not need to define
    intelligence.
    

    View Slide

16. The Turing Test
    How can we tell whether a machine is intelligent?
    ! Don’t we have to define intelligence first?
    ! Don’t we have to “reverse engineer” ourselves to
    build an intelligent machine?
    Turing’s answer: we do not need to define
    intelligence.
    We just need to talk to a machine and not be able
    to tell its answers apart from human answers.
    

    View Slide

17. The Chinese Room
    

    View Slide

18. The Chinese Room
    Searle doesn’t understand Chinese.
    

    View Slide

19. The Chinese Room
    Searle doesn’t understand Chinese.
    The instruction book enables the room to pass the
    Turing Test.
    

    View Slide

20. The Chinese Room
    Searle doesn’t understand Chinese.
    The instruction book enables the room to pass the
    Turing Test.
    Computers function like this.
    

    View Slide

21. The Chinese Room
    Searle doesn’t understand Chinese.
    The instruction book enables the room to pass the
    Turing Test.
    Computers function like this.
    So Artificial Intelligence is impossible.
    

    View Slide

22. Making sense of language
    

    View Slide

23. Making sense of language
    

    View Slide

24. Making sense of language
    Syntax
    

    View Slide

25. Making sense of language
    Syntax
    Rules for transforming symbols.
    

    View Slide

26. Making sense of language
    Syntax
    Rules for transforming symbols.
    “To form the past tense, add an ED to the end of
    the infinitive form of the verb.”
    

    View Slide

27. Making sense of language
    Syntax
    Rules for transforming symbols.
    “To form the past tense, add an ED to the end of
    the infinitive form of the verb.”
    Computers are good at manipulating symbols
    according to rules.
    

    View Slide

28. Making sense of language
    Syntax
    Rules for transforming symbols.
    “To form the past tense, add an ED to the end of
    the infinitive form of the verb.”
    Computers are good at manipulating symbols
    according to rules.
    Semantics
    

    View Slide

29. Making sense of language
    Syntax
    Rules for transforming symbols.
    “To form the past tense, add an ED to the end of
    the infinitive form of the verb.”
    Computers are good at manipulating symbols
    according to rules.
    Semantics
    The meaning of the words we use.
    

    View Slide

30. Making sense of language
    Syntax
    Rules for transforming symbols.
    “To form the past tense, add an ED to the end of
    the infinitive form of the verb.”
    Computers are good at manipulating symbols
    according to rules.
    Semantics
    The meaning of the words we use.
    “Hund” means “perro” in English.
    

    View Slide

31. Making sense of language
    Syntax
    Rules for transforming symbols.
    “To form the past tense, add an ED to the end of
    the infinitive form of the verb.”
    Computers are good at manipulating symbols
    according to rules.
    Semantics
    The meaning of the words we use.
    “Hund” means “perro” in English.
    Can computers grasp meanings?
    

    View Slide