You are on page 1of 3

Can Computers Think?

Over the last few decades, computers have evolved from just being a tool that help in
accomplishing simple tasks. With the advancements in technology, computers can now draw
conclusions and predict the outcomes to solve the most sophisticated problems that even humans
cant solve. Computers have evolved from the tool that is just used to do simple calculations and
manipulate some documents to a tool that you can talk to when you are bored, to a tool that can
autonomously drive for you from one place to the other, to a tool that can win against the best
Chess and Go players in the world. Although computers could solve many of the worlds
sophisticated problems, still computers cant approach the human intelligence in showing
emotions or answering philosophical questions. So, can computers think?

Thinking is the ability to think intelligently about complicated things according to

Merriam Webster. Making the computers acquire an artificial kind of intelligence to solve
complex problems in the world make them great thinkers. The topic of artificial intelligence goes
back to the 1940s after Alan Turing published the paper Computing Machinery and
Intelligence where he questioned whether a computer can think and stated that computers can
learn from experience. Since then, research in Artificial Intelligence went through up and downs
and it wasnt before the 1990s when the research boomed and real products started to appear
that showed some form of intelligence. That was because are two reasons for that. First, the
advancements in the computer hardware technology opened the opportunity to open fields that
requires fast computations and storing large amounts of data that is used in developing the
intelligence of the computer. Also, researchers focused on developing artificial intelligence that
solve specific problems such as image recognition and language translation rather than giving the
machine the ultimate intelligence.

A subfield of Artificial Intelligence called Machine Learning succeeded to create

very intelligent computers by creating a model for the computers that is very similar to the way
the brain works. Machine learning is the process by which the machine learns from experience
and data. It is different from any traditional way of programming. In the traditional way of
programming, the programs let the computer know what decision to make for each possible kind
of input. For each sub problem a new program should be crafted to let the computer know how
to decide based on that type of input. In machine learning, there is a mathematical model that can
be though of as an analogy to the brain. That model is controlled by a set of parameters. As
humans cant make the correct decisions or actions in the beginning of their life because they
have no previous experience, the mathematical model cant produce the correct output as
parameters are still not tuned. As humans grow and gain experience, they gain experience which
affects their thinking flow positively which allows them to take better decisions and become
wiser. That is similar to the training process of the model which tunes the mathematical model
parameters using a previous dataset for which the output is known. After the training process is
complete, this model if trained properly should be able to use any input data to take the correct
decision or to predict the correct output with great accuracy. The whole process of training and
making decisions using the machine learning approach is very similar to the way the brain works
and thinks giving a great evidence that computers can think.

There are many mathematical models that can act as the brain of the computer. Examples
for these models are decision trees that take a group of successive decisions that leads to a
certain output, support vector machines in which kernels are used to transform data to separate
them into categories and neural networks that actually imitates the physical way in which the
brain works. It consists of computational units called neurons that combine some inputs to
produce a certain output. Layers of neurons are connected to each other to form what is called a
deep neural network. Deep neural networks allow the model to detect successive patterns in the
input data to be able to make the correct decision. For example, in image recognition, pixels are
the raw input data for the neural network. The first layer of neurons detected some patterns. The
second layer detects patterns of patterns. Each layer uses the patterns detected in the previous
layer to produce more patterns which are used to produce the final output. For each neuron, there
is an associated weight which can be though of as the importance of a certain input or pattern
relative to the others. The learning process for neural networks is the process by which those
weights are tuned using a known data set. So the process by which the neural network learns is
the same for all kinds of problems. What differs a model from the other is the the architecture of
the neural network and the final weights on the neurons. The research in that area has led to to
the emergence of neural networks that can solve many of the problems we face with huge
complexity. Deep learning has closed the gap between the brain of living organisms and the
computers proving that computers can think if the problem can be broken down to mathematical
basis and we have enough data to train our models.

Machine learning in general and deep learning specifically have led to great computing
products that can think logically and solve many of the complex problems we face daily. Now in
2016 many companies are in the race of releasing full autonomous cars in the market. A car by
tesla recorded 130 million miles driving successfully autonomously before it crashed which is a
great record. Many companies are releasing chat bots that act as personal assistance like Siri by
Apple and Allo by Google. They are bots that can chat with you and uses data from the internet
to answer you questions and even do some tasks for you such as setting a reminder for you and
reserving a table in a restaurant.

Computers can think in their way by doing mathematical operations that can lead to the
emergence of many computing devices that help as with our daily lives. However, there are other
types of thinking that cannot be represented in terms of math such as emotional thinking. In the
end, computers are mechanisms not organisms. They do not have feelings. It is very hard for a
computer to acquire the emotional artificial intelligence that makes it love or feel sympathy for
others. Also, computers cant think philosophically. Computers still dont have the ability to
have the philosophical reasoning to answer questions like Why do we exist? and Who created
us? because they cant be simply broken down into a mathematical problem that can be solved
using a series of calculations.

ToDo: Conclusion