Artificial Intelligence: The second sixth day - Part 1

On The Sixth day man created . . . AI

If you are a Christian (like me) or a Muslim or a Jew you believe that on the 6th day of creation God created Man - the first man. This first man was a notch above all other creature's - made in the image of God himself.

'Man' and his companion 'Woman' were the epitome of Gods creation, given dominion over all creation - animate and inanimate. Together they formed Humanity.

Whether you believe this or not, you will agree that humanity has studied and tried to explain the world he lives in through creation of the distinct Natural Sciences - Mathematics, Physics, Chemistry and Biology.

There is one other science though. It grew out of the first four sciences and represents a fundamental shift in the paradigm of science - from the search for truth and understanding to creation of the truth.                                                This new science is Computer Science - and the epitome of the paradigm shift it represents is the sub-discipline of Artificial Intelligence.

 - With Artificial Intelligence humanity is trying to re-enact his creation -

Artificial Intelligence is the study of the creation of non-natural (not humans, other animals or plants) intelligent agents that are capable perceiving their environment, of figuring out and taking actions that maximize the chances of successfully achieving their goals.

An example would be an 'intelligent' robot that is placed in a maze and must find the exit. The program must realize that its in a maze, it must learn that walking into a wall is not a good idea, it must realize that the walls are to high and slippery to climb and it must figure out that some turns lead to dead ends or loop back. Most importantly our intelligent robot must find the exit, faster and with fewer mistakes than any human being would. All this without being 'programmed' to navigate mazes but rather being programmed to use reason, logic and learning to figure things out.

AI is the result of hundreds of years of human imagination and decades of scientific research.

In 1956, a group of eminent American Mathematicians met for eight weeks at Dartmouth College in New Hampshire for a loose workshop/extended brainstorming session taking the form of presentations and discussions. Topics like Natural Language Processing, Neural Networks and Abstraction discussed at this early AI camp went on to become fields of research and finally mainstream technologies.

John McCarthy - organizer of the 1956 Dartmouth Workshop

This was a group of really early pioneers (there were many others around the world) and the concepts they discussed and went on to research would take the better part of half a century to reach main stream computing.

As computing and AI became more popular, what we thought AI would turn out to be and what we actually got seemed a bit different. As evidenced by the Turing Test, the early computing community believed the goal of AI was to create agents that mimicked us [humans] as closely as possible.

This very narrow definition and application proved to be a bar to far in the early days of computing and instead successes were found in less glamorous domains, like playing Tic-Tac-Toe (a.k.a X & O), chess, crossword puzzles, facial recognition, optical character recognition, video games, limited domain robotics and mathematics.

As is typical - early research successes in these unglamorous fields have proved useful in building blocks and learning experiences in more complex applications of AI.

But what really makes up AI? What are the components of an Intelligent Agent?

• Ability to Reason & Problem Solving: Early researchers developed algorithms that imitated step-by-step reasoning that humans use when they solve puzzles or make logical deductions. More recent research particularly in neural networks attempts to imitate the human brains functions that enable us to make intuitive judgments.
• Knowledge Representation: all naturally intelligent organisms are aware of millions of atomic facts - like what water feels like to touch, what a tree looks like, the effect of gravity etc etc. AI agents need to be able to store and represent knowledge in a structured way that allows for categorization, building relationships between sets of data, aggregating it etc. This knowledge can then be used in problem solving.
• Environmental Perception: in order to solve a problem, an agent must be aware of its operating environment, the variables in it and how they change as the agent acts in it. Computer vision research has matured to the extent that industrial and commercial applications are now common place.
• The ability to Plan:  given a goal, an agent must be able to layout a set or sequence of actions it must perform to achieve the goal. It must also pick the most optimal set of actions and be aware of how each of those actions get it closer to the goal.
• Machine Learning: experience is the best teacher - only if you learn from it. AI agents must be able to assimilate the experience of others and its own experiences. Determine what worked well or did not and what can be done better and how. This field of researched has reached come of age recently and is seeing increasing application in everyday life.

Do you know any more critical components of AI? Drop us a comment :)