Can/could machines be thinking things?
To start talking about whether machines can/could be
thinking things, it is first required that I explain what I mean by ‘thinking’,
how do we as humans think, and ‘machine’, what type of machines would be potential
candidates for thinking. However, I also
need to discuss whether thinking is intelligent as to see whether the machine
is then an intelligent machine. Then I will bring the two definitions together and
try and answer the question from a view of dualism, functionalism and materialism
to see how they view thinking machines and whether the answer differs from one
view to another.
What is ‘Thinking’?
We have long seen ourselves (humans) as the top of the food
chain and animal kingdom and this is partly due to our brains. Not entirely the
physical side of them but more the power they possess to ‘think’. So, what is
thinking? Thinking has been shown in books or comics as the ability to speak to
ourselves with in our ‘mind’. Many books have shown thinking by having the
character talk to themselves in their brain but without speech marks which
indicates that it isn’t being spoken aloud. Comics usually portray thinking as
a bubble above the characters head which has words of their inner thoughts. This
shows that humans commonly refer to thinking as some form of internal
discussion with ourselves as this is what our experience of thinking seems to
be (for the most part). This suggests that thinking requires an understanding
of a language, to have some sort of internal dialogue. This would show that
thinking is then reliant on a language to be able to happen. Charles Fernyhough
suggests that thinking isn’t one thing that we do, instead he claims that “There
are (at least) two kinds of inner speech, what I have called condensed and expanded” (Fernyhough,
C. (2004). p49-68). He claims that condensed inner speech is still a language,
but it is not spoken as its acoustic properties have been removed. The other
form of thinking he said was expanded, which is where you have a proper
internal dialogue with yourself, as if you were chatting to someone.
Elizabeth Spelke, cognitive psychologist at Harvard university,
argued against Fernyhough by saying that she often has conscious thoughts that
have no words, to which Fernyhough countered by giving two reasons, the first
being that thinking isn’t a conscious thing and the second stating that the
experience she described is his first type of thinking, condensed inner speech.
Is ‘Thinking’ intelligent?
Machine intelligence often requires the machine to ‘think’
about what it is doing to overcome its problem, like that of humans but the
question arises to whether thinking is intelligent. The question itself could
be answered by stating if a machine acts as intelligently as a human then it is
an intelligent machine and if a machine is required to think, then thinking is
linked to intelligence. Alan Turing decided instead of asking whether a machine
can think, he would ask the question in a different form by creating what he
called the Imitation game. This game would be where you have 2 people and a machine,
one person is having a conversation with a computer and one who is an
interrogator is to determine which one is a machine, and which one is human. All
participants are separated, and the interrogator can view the conversation
between the human and the machine. If he can’t find out which one is a machine,
then it is said that the machine is showing intelligent behaviour due to the
machine’s conversation with the human fooling the interrogator into whether it
is the machine or not. Turing believed that for a computer to do well in this
game, indeed fool the interrogator, then the machine would pass his test and
answer the question “can machines think”. Turing decided the best way for the
machine to play the game would be to make the machine give answers like that a
human would give, as this would mean the interrogator can’t tell which is which
due to both participants acting very similar in the conversation. He did
however say that in the future it may not be the best case for a machine to
recreate a human’s intelligence but for the time being it satisfied his answer as
we class ourselves as intelligent.
Turing did get some objections against his view, such as the
theological objection for which dualists claim that thinking is part of the
immortal soul, and as machines do not have souls, they cannot think. It may
even go as far as to say that to have a soul in one’s body, it must have been given
to them by a divine creator (e.g. God). As it is up to God on whether the body
has a soul, we have no power to gift one to a machine for it to think. However,
this argument is not held by many because there are a lot of people who object
to theism and do not believe in a divine creator. Even if there was a god,
there is not much support for why god wouldn’t give a machine a soul to be a
What is a ‘Machine’?
Machines come in many different forms, Alan turning used the
definition of machine as a digital computer for this question when he created
the the imitation game. Turing Stated that “a digital computer can be usually regarded
as consisting of three main parts” (Turing 1950, p437) which where a storage
place, an executive unit and a control unit. The storage place was where
information was held, so all things you needed to hold onto are stored here. The
executive unit is the unit which computes all the operations done by the
computer itself. These operations vary in difficulty for the computer, from simple
tasks such as display this number to repeating a function multiple times over
until a desired outcome has been achieved. The control unit is a manager of
sorts, it maintains the order of operations, keeping everything in track. The machine
deals with packets of information at a time, so one packet may give a certain
command that adds up numbers from a certain location in the computers memory
(storage place) to another in a different location. All these packets come in
the form of numbers as computers and not in English, so the computer can deal
with operations from a table without having to deal with the semantics of
words. This definition of a digital computer gives us an understanding of the
type of machine that will deal with this question.
The most obvious example of a digital computer is that of
the common desktop computer or laptop that a clear majority of humans use around
the world daily. However, there are machines that currently fall inside the category
of digital computers which may at first seem intelligent, such as the most
recent car technology, autopilot, produced by a company named Tesla owned by
Elon Musk. The car which inhibits this technology is made up of multiple
sensors that sense the environment around that car to then steer it without human
interaction. This may seem intelligent at first but it can only do a very
minute task in which a lot of humans do every day, and it lacks the ability to
do anything outside of controlling the car.
Can/Could Machines Think?
Now we have our two definitions defined, and whether it is
intelligent, I will attempt to answer this question from more than just one philosophical
view. If one is to claim to be a dualist then they believe that the mind and
body are separate. In this instance, being a dualist, you could argue that the
answer would be no because you cannot recreate a mind without a soul and assuming
we need a mind to think as the mind leads to mental states which give way to
thought, then creating a machine to have the ability to think would be impossible
unless we could recreate a soul, which we cannot (as explained in ‘is Thinking
intelligent?’). Dualism also states that the machine would have to be conscious
that would mean having to define what consciousness is to see why a machine
cannot be conscious. However, defining what consciousness is a hard task as it’s
not something humans have yet fully grasped but as it is fundamental for understanding
the mind, I cannot give an argument for consciousness.
Another philosophical view is functionalism, which John R
Searle gives an argument against machines being able to think, in the form of a
thought experiment “The Chinese Room” where a person is put in a room with a book
of rules and a paper of Chinese questions and a bunch of paper. The experiment
dictates that if the user was to simply follow the rule book and give an answer
to each question by using the rule book, they would not have an understanding
for Chinese. If we replaced the person with a machine and the rule book with a program
for which the machine has been programmed with, we conclude that machines don’t
understand what they are doing as they lack semantics and the machine would
just be a going through a simulation and nothing more. This suggests that if
someone was to try and create a machine that could think it could likely end up
being more of a simulation of thinking and being an intelligent machine rather
than thinking itself. Seeing as we have yet the ability to simulate human
intelligence, it is hard to argue with Searle, although his argument could be
used as a test for the Turing’s imitation game as it is hard to prove that a
computer was simulating a human without being able to test it out.
Materialism (also known as physicalism) is the philosophical
view that says everything is physical in some form. Materialists might say in
the future it will be possible to give the machines the ability to think
because if everything is physical then we can recreate the necessary human
functions that make us think and place them in a machine. Although this cannot
be done as of right now, since we lack the technology to create a machine in
which we can give it the ability to think, Eugene Izhikevich gave a simulation
of a human brain ( neurons) and almost one quadrillion () synapses in 2005 and
it took 50 days with 27 processors, running at 3 Gigahertz each, to simulate one
second. Estimates have been made for when humans might create the first machine
that can think, one estimate was made by Ray Kurzweil in his book titled ‘The
singularity is near’ where he predicted that in 2029, humans would have created
a machine capable of exhibiting human intelligence that passes the Turing test.
After all that I have spoken about in this essay, I conclude
that thinking is formed from two types of internal speech which both rely on language
in some form, that intelligence is related to thinking and intelligence in a
machine can be proved by using the imitation game as a test for intelligence. I
also see dualism as disagreeing with the question on whether machines can or
cannot think due to the lack of a soul. Functionalism also argues against the question
on whether machines can or could think and uses John R Searles thought
experiment, the Chinese room, to show that we are more likely to simulate thinking
rather than have the machine think for itself. Finally, Materialism partially
agrees and disagrees with the question as it isn’t currently possible to create
a machine that is able to exhibit human intelligence. I more aligned with Materialism
argument in that we may one day create a machine capable of exhibiting human intelligence.
(word count 1985)
Campbell, T. (1997). Thinking About Thinking | Issue 18
| Philosophy Now. online Philosophynow.org. Available at:
https://philosophynow.org/issues/18/Thinking_About_Thinking Accessed 8 Jan.
Cole, D. (2004). The Chinese Room Argument. online
Plato.stanford.edu. Available at:
https://plato.stanford.edu/entries/chinese-room/ Accessed 8 Jan. 2018.
En.wikipedia.org. (n.d.). Philosophy of artificial
intelligence. online Available at: https://en.wikipedia.org/wiki/Philosophy_of_artificial_intelligence#Intelligence
En.wikipedia.org. (n.d.). Turing test. online
Available at: https://en.wikipedia.org/wiki/Turing_test
Fernyhough, C. (2010). What do we mean by ‘thinking’?.
online Psychology Today. Available at:
Izhikevich, E. (2005). Eugene M. Izhikevich,
Large-Scale Simulation of the Human Brain. online Izhikevich.org. Available
Kurzweil, R. (2005). The singularity is near. p.167.
Plato.stanford.edu. (2001). The Turing Test (Stanford
Encyclopedia of Philosophy). online Available at:
Plato.stanford.edu. (2004). Consciousness (Stanford
Encyclopedia of Philosophy). online Available at:
Plato.stanford.edu. (2003). Dualism (Stanford
Encyclopedia of Philosophy). online Available at:
Thompson, C. (2016). Here’s how Tesla’s Autopilot works.
online Business Insider. Available at:
TURING, A. (1950). I.—COMPUTING MACHINERY AND
INTELLIGENCE. Mind, LIX(236), pp.433-460.
Fernyhough, C. (2004). Alien voices and inner dialogue:
Towards a developmental account of auditory verbal hallucinations. New
Ideas in Psychology, 22, 49-68.