BUILDING BLOCKS OF ARTIFICIAL INTELLIGENCE

INTRODUCTION

We are living a new technological revolution, a revolution that will transform our lifestyles drastically, a revolution caused by the advent of Artificial Intelligence (AI).

Today, AI is equated with killer robots itching to destroy the human race, an idea bred by Hollywood movies. But AI is more than that. Broadly speaking, AI’s objective is to build intelligent entities, such as machines or software, to facilitate our daily tasks and to bring comfort to our lives. John McCarthy, who coined the term in 1955, defines it as, “It is the science and engineering of making intelligent machines, especially intelligent computer programs.” Numerous other rigorous definitions have been provided, but a single universally accepted definition of AI is yet to be established.

THE FOUNDATIONS OF AI

It is evident that to build an intelligent machine, we have to first understand intelligence. Therefore, AI is a multidisciplinary field, founded by ideas from numerous other subjects.

Philosophy

The question that bothers me the most is: does the mind create our thoughts, or is it our brain? Put differently, if we could recreate a brain exactly, would it function like any other brain? Would it display consciousness, will and creativity? Further, if consciousness is not the effect of the wiring in our brain, but in fact the cause of our thoughts, how do we create consciousness?

Following this stream of thought, other questions arise: what does it mean to understand something? Is logic inherent? What is knowledge? Does it differ from information?

From the times of the ancient Greeks, people have been troubled with the functioning of the mind, reason and logic. Aristotle’s theory of Syllogism describes the basic process of the rational mind, where we deduce conclusions from an initial premise. He says that:

A deduction is speech (logos) in which, certain things having been supposed, something different from those supposed results of necessity because of their being so. (Prior Analytics I.2, 24b18–20)

This is nothing but the notion of logical consequence or logical implication, which is the base of mathematics. More information on syllogism is available here.

Syllogisms describe the workings of the mind, but what is the mind? René Descartes offered a theory saying that the mind is a substance whose essence is thought. Further, mind and body are distinct, a theory known as the “mind-body dualism“. Further reading can be done here.

Materialism holds an alternative view and proposes that the mind is merely the result of the interaction of matter, a view that seems rather narrow.

We have discussed the workings and the substance of the mind, but what about consciousness? Is consciousness a substance, or merely a result of the interaction of matter. Sri Aurobindo Ghosh posits that consciousness is the fundamental substance of this universe, and is involved in every material object. Hence, evolution is simply the effect of the emergence of consciousness. His major work, The Life Divine, explains his philosophy thoroughly.

I have merely scratched the tip of the iceberg of the philosophical theories which found AI, but it is enough to make you aware of the complexities of AI.

Mathematics

Philosophy delves in the realm of ideas. But to make the ideas concrete, a formal set of laws, logic and notations are required. And this is where comes mathematics. Though approaches to the subject vary from researcher to research, in general here are the topics used:

-Logic : propositional, first-order and fuzzy.

-Linear Algebra

-Probability Theory

-Statistics

-Algorithms

-Calculus

-Optimization Theory

-Graph Theory

A strong mathematical foundation is a must for AI, for without it, you’ll be floundering in the ocean of jargon and symbols.

Computer Science

There are two sides of computer science that needs to be developed to create AI: hardware and software.

A basic background of the architecture of computers is necessary to be able to create a functional intelligent program and to understand the flaws and restrictions. In the future, hardware too can be modified to be more befitting to our aim. What if we create a computer shaped like a brain? Would it perform faster? What tasks would be easier? What would be the restrictions? Could we create artificial neurons?

AI is finally a computer program, a software. Therefore, a thorough understanding of data structures, algorithms, computer networks, databases and programming, especially object-oriented programming, are a must.

There are tons of programming languages out there, but currently, here are the top ones used for AI/ Machine Learning/ Data Mining/ Data Science/ Analytics.

ml2_graph

Source: Languages and Libraries for Machine Learning

Further information on programming languages can be obtained here.

Neuroscience

To create intelligence, it is obvious that we have to first understand how our brain functions. This task is accomplished by neuroscience. Which areas of the brain work we reason? think? imagine? see? hear? How is information stored in the brain? What creates thoughts? What happens when we dream?

I will elaborate on the studies of neuroscience in a later post.

Psychology

Neuroscience studies the physical functioning of the brain, but what about our behaviour? How do we act? How do we make decisions? How do we reason?

Cognitive psychology studies the mental processes behind these actions, and further attempts to give a theory to the functioning of the brain. Subsequently, taking inspiration from these theories, we can create AI.

Linguistics

Finally, humans have to interact with machines. Therefore, the machines have to understand our language and all the underlying nuances. This task at a first glance doesn’t seem too difficult, but is actually quite complex. To understand a sentence, understanding the grammar is not enough, one has to understand the context and the matter. How do you make a machine understand the concept of a “cat”? What about an abstract concept such as “love”? You could provide a definition, but would it understand? The difficulty, as you might realise, is not the syntax, but the semantics.

This gives only a glimpse into the fields that are at the foundations of AI, but it is enough to reveal the magnitude of the difficulties ahead. In the following post, I will explore the tasks that AI aims to accomplish.