AI Flashcards
What is the best test for AI?
The Turing test
What is AI?
A research program in computer science that aims to create computers which display ‘intelligent’ behaviour
How is the mind like software?
Functionalism recognises the mind as ‘neural software’. Just as the functions of computer hardware cause software to run, the functional organisation of the brain leads to consciousness.
How does functionalism support AI
Mental states are multiply realisable, so could be realised by a computer system. There is nothing, theoretically, which would prevent the construction of a machine functionally isomorphic to a human being, which would therefore be conscious
What does the computational theory of mind claim about thought?
It claims that thought, like computation, is simply symbol manipulation.
How does Fodor’s language of thought hypothesis support CTM?
It suggests that thought is a structured language, and therefore consists of structured symbol manipulation.
What properties does thought possess, which structured languages also do?
Productivity and systematicity
Why wouldn’t an unstructured language have systematicity?
There would not necessarily be a symbol for ‘X did Y to Z’ as well as ‘Y did X to Z’
First problem with AI
The Chinese room argument
What is the point of the Chinese room argument?
The responses produced by the system are meaningless, because they do not possess intentionality. Intentionality cannot be derived purely from functions.
What is intentionality?
An intentional state is a mental state which is about, or directed upon, a representative mental content. For example, my belief that ‘London is in England’ is about London; it represents London as being in England.
What does a belief exemplify?
Russell’s propositional attitude
What are the three features of a propositional attitude
Representative content, ‘that’ clause, and cognitive stance
What are the 2 replies to the Chinese room argument?
Systems reply and robot reply
How does Searle reply to the systems reply?
The man in the room could theoretically internalise the entire system - still couldn’t derive semantics from syntax!
