Could a Machine Think ? LILY-MAY EDWARDS
Artificial Intelligence ( AI ) has rapidly developed over recent years , these extraordinary machines which have been created has left us pondering over many questions . One of these is could a machine think ? I will be arguing that it isn ' t possible for a machine to think , due to these machines not having consciousness . I will argue that thinking requires consciousness , therefore if a machine is not conscious then it cannot think . I will begin by describing the unconvincing view of Strong AI , then I will discuss the ' Turing Test ' and its effective test for intelligence , but not effective for testing if a machine is capable of thinking . I will conclude with Searle ' s ' Chinese Thought Experiment ' and its idea that there is more to human thinking than just intelligence .
Strong AI is the view that AI demonstrates human-level intelligence . Supporters of strong AI essentially argue that machines can behave , think and understand the same way a human can ( Mandik , 2014 ). The development of machine learning is a good argument in favour of strong AI . Machine learning algorithms allow machines to learn from data and past experiences , essentially programming themselves to produce more effective outcomes ( Hauser , n . d .). Some may argue that machine learning can be seen as evidence of the machines ' thinking ' because it shows the machines understanding from their past experience and rectifying it . I disagree with this argument and instead believe it is purely intelligence , as they aren ' t thinking as such but are using their intelligence more effectively . I don ' t believe that machines are capable of thinking like the proponent of Strong AI would suggest . Instead I think that machines are just highly intelligent and have been programmed to imitate humans and their behaviour .
Alan Turing posits the infamous ' Turing Test ' to test for intelligence among machines and to see if they are capable of thinking like a human . Turing proposed that a machine can be said to be intelligent if it can imitate human behaviour ( Hauser , n . d .). For the test to be effective , intelligence should be defined as ' producing an appropriate response in a variety of situations '. It ' s reasonable for us to use this definition because although it is a broad definition , I accept this as we already talk about the machines we have . Using the term AI , we seem to be content using our traditional concept of intelligence to talk about lots of existing technology . However , I don ' t think the test proves machines have the ability to think , there is more to thinking than just intelligence . Thinking requires being able to understand , which requires consciousness and machines don ' t possess this consciousness . Ultimately , the Turning Test can be seen as being effective in testing for intelligence , but not effective when testing for the ability to think .
Even if a machine can pass the Turing Test , it still wouldn ' t be able to think . It could be argued that to truly test whether or not a machine can think , we need not to focus on the outward behaviour like in the Turing Test , but instead focus on the internal cognitive processes and subjective experiences . This is demonstrated in Searle ' s ' Chinese Room Thought Experiment ' ( Searle , 1980 ). In Searle ' s thought experiment , a person is placed alone in a room . In the room there is a
LILY-MAY EDWARDS
161