Artificial Intelligence (AI)
What is Artificial Intelligence? What Do We Have Achieved So Far?
Artificial Intelligence
Artificial Intelligence (AI) is extensive branch of computer science concerned with building intelligent machines that mimic human actions and capable of performing tasks that usually require human intelligence. AI is a multi-disciplinary science with multiple approaches based in psychology, linguistics and computer science which endeavor to simulate human intelligence in machines.
Goal of Artificial Intelligence
Alan Turing’s (father of modern computers) Turing test established a fundamental goal of artificial intelligence which states that
“A computer would deserve to be called intelligent if it could deceive a human into believing it was human”.
Expansive goal of Artificial Intelligence makes it very difficult to explain. This is the reason that no single definition of Artificial Intelligence is universally acceptable. Simply said an intelligent machine with human like qualities may be referred as AI. However, one might wonder what makes machines Intelligent?
In an attempt to answer the question authors Stuart Russell and Peter Norvig explored four different approaches in their groundbreaking textbook Artificial Intelligence: A Modern Approach that have historically defined the field of AI:
· Thinking humanly
· Thinking rationally
· Acting humanly
· Acting rationally
First two ideas concern thought process — ability of any machine to calculate the best possible result and rationalize the output with the given data. The other two ideas concern behavior — such that human intelligence is defined in a way that it can be easily simulated in machines so that they can execute tasks much alike humans.
These definitions may seem abstract to the average person however, they give a basic idea about the objective of AI.
Types of AI
Generally, Artificial Intelligence is divided into two types:
· Artificial Narrow Intelligence (ANI)
· Artificial General Intelligence (AGI)
Artificial Narrow Intelligence (ANI)
Artificial Narrow Intelligence (ANI), sometimes referred to as weak AI is a type of AI that we see all around us on daily day basis. ANI is realization of our dream of building humanoid robots so far. It is usually designed to perform one particular task. ANI has experienced numerous breakthroughs in the last decade and created a lot of hype on social media. These machines may carry out some tasks better than humans but operates under far more constraints than basic human intelligent. For Example, to make a machine able to recognize a human face or a car or a cat using ANI requires multiple iterations however, this is not necessary in the case of humans. It is because we are not yet able to develop an AI that can remember the choices it made in the past in a certain scenario and act likewise in that same scenario in the future rather than doing the same calculations all over again.
Applications of Artificial Narrow Intelligence (ANI)
This is what we have accomplished with ANI so far:
· Email Spam Filter
· Image recognition software
· Self-driving cars
· Virtual Assistants (Siri, Google Assistant etc.)
Working of ANI
Much of the breakthroughs in ANI today are made possible with breakthroughs in Machine Learning and Deep Learning.
Machine learning can be defined as:
“Field of study that gives computers the ability to learn without being explicitly programmed” Arthur Samuel (1959)
Deep learning, on the other hand, is a network of artificial neurons which processes data and perform certain operations on it where a neuron or artificial neuron is a set of functionalities. Each artificial neuron in the neural network is specified to do a different task. Deep learning was also referred to as neural network or artificial neural network because it was assumed that artificial neural network is similar to the neural network in brain, however, there is no similarity between them at all.
Understanding the difference between artificial intelligence, machine learning and deep learning can be confusing. However, Frank Chen provides a good overview of how to distinguish between them, noting:
“Artificial intelligence is a set of algorithms and intelligence to try to mimic human intelligence. Machine learning is one of them, and deep learning is one of those machine learning techniques.”
Simply put, Machine learning is a subset of artificial intelligence and deep learning is a subset of machine learning.
Machine learning feeds data to computer and uses techniques such as deep learning to get gradually better at a task without being explicitly programmed.
Artificial General Intelligence (AGI)
Artificial General Intelligence (AGI), also known as strong AI, is the kind of artificial intelligence we see in the movies, like the robots from I, Robot and Westworld. AGI is capable of performing wide-range of tasks much like human beings. It has General Intelligence like humans and it can apply that intelligence to solve almost any problem. In short, a machine with a full set of cognitive abilities. We are too far from building these machines yet and for now they can only be seen in the movies. AGI has long been the muse of dystopian science fiction in which super intelligent robots wreak havoc on Earth.
What Do We Have Achieved So Far?
Since its beginning, AI is over hyped by public and researchers alike, idea that machines will become so intelligent that they will take off on their own and humans will not be able to keep up with them. However, the truth is, at this point in time, we are only able to make smart speakers, chat-bots, virtual assistants and all those applications that are powered by artificial narrow intelligence. We are very far away from building machines with human-level intelligence which is a dream of many people.
The new White House report on artificial intelligence vividly describes our development in the field of AI to date. It says the next 20 years likely won’t see machines “exhibit broadly-applicable intelligence comparable to or exceeding that of humans,” though it does go on to say that in the coming years, “machines will reach and exceed human performance on more and more tasks.”
There is a misconception that Artificial Intelligence is all about building humanoid robots, however, researchers argue that maybe we don’t need humanoid robots. Perhaps, the more important obstacle to overcome is not the physicality, it is making computers to learn.
