Can machines think? (Turing, 1950)



1. Can machines think?

Defining the words ‘machine’ and ‘think’ could be dangerous and make us think the answer could be found in some form of statistical survey

A = Making C make the incorrect identification

B = Making c make the correct identification

Both could say: Don’t listen to him

→ There is little point for the machine to appear human if we want to measure intelligence

→ What is a machine in this context? A ‘digital computer’

Digital computers

Human computers → Following a set of fixed rule using an unlimited supply of paper for calculations

Digital computers → Old idea from Charles Babbage

• Store of information for calculations and rules‘Add the number stored in position 6809 to that in 4302 and put the result back into the latter storage position’Would be expressed as 6809 4302 17Where 17 references the type of instruction defined in the rulebook
• Excecutive unit Which carries out individual operations
• Control which ensures that instructions are performed in accordance with the rules and in the correct order

Programming → Creating rules that computers must follow

Contrary views

Theological objection

Argument: We think because we have soul which is not present in animals or machines therefore they can’t think

Response: Machines would usurp god’s power of creating souls any more than humans do so with children

“Artificial Intelligence began with an ancient wish to forge the gods” - Pamela McCorduck

• The fact that humans create consciousness by having children

Head in the sand objection

Argument: Let’s hope machines aren’t dreadful

Response: Intellectuals might be biased to this argument since they value thinking more than others

• What?

Mathematical objection

Argument: Discrete-state machines have limitations (IE Godel’s theorem) (Any sufficently good logical system there can be statements that can be proved or disproved

• Elaborate on godel’s incompleteness theorem

Response: Humans are also bounded to these mathematical limitations yet we exist

Consciousness argument

Argument: It’s not intelligent since it’s not conscious

Response: Solipsim is an eternal loop, so the chances is that we all think, it’s true that consciousness is mysterious but doesn’t have an impact on this question (Why?)

Various disabilities

Argument: Machines may do all those things but will never be able to do X

Response: Induction (past machines have been clunky therefore they must stay clunky) + Anthropocentric bias (We pretty much project ourselves) = Function diversity is a consequence of storage which will increase exponencially

Ada Lovelace

Argument: Babbage machine did not have originate any behaviour on it’s own

Response: Doesn’t imply impossibility

• Study their dynamic, how did lovelace learnt from babbage?

Continuity in the nervous system

Argument: NS is not discrete, small chemical differences have big effects in the outgoing impulse, one cannot replicate the same effect with a discrete system machine

Response: True but the interrogator cannot take advantage of this fact, it would be possible to mimic continuous outputs

Informality of behaviour

Argument: We have some rules but our common sense adds infinetly more rules and makes it impossible to make discrete

+ Commonsense project

Response: We clearely follow behaviour norms, conduct rules are hard but not impossible, there’s no point where we say ‘we haven’t found those laws, everyone is fundamentally different’

Extrasensory perception

Argument: There might be thelepathy and other powers

Response: We can adapt the room for thelepathy-proof room

Intresting how he considered the position was strong

Learning machines

Turing’s main points:

• Mind = Onion, which layers makes up the onion?
• We need to wait until the end of the century and plying the imitation game
• This is a programming problemThe brain might be from 10^10 to 10^15 binary digits. Most of it used for visual information + How powerful is the brain?I’d be surprising if more than 10^9 would be required for playing the imitation game since little areas are devoted to higher level thinking
• Adult mind: birth, education, additional experienceSeparate the problem into two, simulate evolution
• Create brain
• Educate it
• Varying complexity
• Order of decisions matter → Difference between a brilliand and poor reasoner
• Intelligent behavior is no longer disciplined, self-contained predictable computation, ergo harder to understand the pupil
• Random(Evolution) > Systematic(Not possible in the physical world)

Notes

Reframing the intelligence from perception of intelligence

Chinese room → Key distincion from you being conscious and being able to understand something

Instrumental goals → The goals that come up in order for you to archieve your main goal