PHIL 212 Lecture Notes - Lecture 5: Turing Test, Memory Bank, Thought Experiment
![](https://new-preview-html.oneclass.com/aApVWe82G41nmxLnKKr5myZrlzDKwqd5/bg1.png)
Philosophy of Mind
9.26.17 Lecture Notes – Artificial Intelligence
Midterm: 4 MC; 4-5 short answer – ask you to explain a concept or theory (e.g., what is LB and
what is 1 problem with LB) – 3-5 sentences; 2 long answer – couple paragraphs, maybe
explaining the thought experiment and evaluate it
Turing, Computing Machinery and Intelligence
- “tats ith the uestio: Ca ahies thik?, ut eliees the uestio is too ague
o Also believes that machines could encompass humans
- Replaes ith Ca e construct a machine (specifically a digital computer) that can
elialy pass the Tuig test?
- For our purpose, a digital computer is a device that:
o Receive inputs, and
o Generates corresponding outputs by following a specific set of instructions (i.e.,
its program)
o Turing talks about other stuff that computers have
▪ Memory bank – stores information
▪ Executive unit – carries out the computing
▪ Control – ensures that the executive unit is following the program
- The Imitation Game
o Iteogato ho a’t see the man and woman, man and woman – through
Q&A, the interrogator is trying to find out which of the two is the man or woman
- The Turing Test
o Rather than having a man and woman, you have a person and a computer
o Interrogator must determine which room contains the person and which
contains the computer
o The person is trying to be a human and the computer is trying to impersonate a
human
o Is the test sufficient?
▪ Could’t the opute iitate hua esposes ithout atually haig
intelligence? – Turing would say the test is sufficient
▪ Also depends on the interrogator
▪ What matters is the results – performing well in the test – kind of a
functionalist approach
o Another gray area – how many times does the computer have to fool the
interrogator?
▪ Turing makes the empirical prediction that computers in the future will
be able to fool interrogations about 30% of the time (p. 442)
- Some bad objections to the possibility of AI
- 1) Computers will never truly be capable of thought because they necessarily lack
consciousness
o T: Why not? Maybe a sufficiently complex computer could have conscious states.
find more resources at oneclass.com
find more resources at oneclass.com