AI or Artificial Intelligence is defined as the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. In simple, AI is the concept of a computer which would have the brain power of human i.e. it can think, reason, react etc.
This idea of building a machine or developing a system to simulate the human brain was originally proposed by John McCarthy, a young assistant professor of mathematics in September 1955. His idea was that this system would do everything a human brain does with intelligence and also learn from its mistakes and improve. He called this new field of study as “Artificial Intelligence”.
John McCarthy’s proposal in 1955 has met with partial success in today’s IT world, more than 60 years later. Today, advances in computers and machines have helped us automate many tasks that were manual and cognitive in nature.
One of the challenges AI faces is its broad term is that some technologists argue that any cognitive technology would classify as falling under AI whereas others argue that AI is a technology in its own right. So which is true? I leave the decision up to you, dear reader.
Coming back to our main topic, let us look at what are the major components of AI. So how do we define Intelligence? The Oxford Dictionary defines Intelligence as “the ability to acquire and apply skills.” Another definition in this dictionary says “a person with the ability to acquire and apply skills”.
A psychologist would define intelligence as a combination of traits or abilities like capacity for learning, ability to apply what has been learnt, reasoning etc. where each of these is an individual trait. Since all the abilities of human intelligence cannot as yet be applied to AI, research has concentrated on learning, reasoning, perception, using language and problem solving as the main focus areas.
Let us take a quick look at these components below:
This is one of the fundamental components of AI learning models, classified into 2 types:
- Feedback based: This is further classified into Supervised, Unsupervised and Reinforced. Supervised Learning is similar to teaching young children where there is a “teacher” to guide the learning process. There are series of sample inputs given with the desired output.
- Reinforced Learning is similar to training an animal and follows the concept of “teach and reward”.
- Unsupervised Learning is the most difficult of these 3 learning models as the machine has to learn in an unsupervised manner similar to a trial and error method.
- Knowledge based: This is further classified into Inductive and Deductive.
To reason means to infer something from the given information. This is also a type of learning as the information is used to reason out a possible path which is relevant to solving a particular problem or situation. Reasoning can be inductive reasoning or deductive reason.
- In Inductive Reasoning, previous similar situations are considered and an inference is drawn based on these considerations.
- In Deductive Reasoning, the situation itself becomes the basis of the inference. This reasoning follows the process of elimination to reach a conclusion. We commonly use Deductive Reasoning in Mathematics.
Problem solving can be defined as systematically searching through possible solutions to solve a particular problem or reach a particular goal. The techniques involved can be categorized as special – purpose and general – purpose techniques.
- A special-purpose method or technique is for a specific problem and targets very specific features of that problem.
- A general-purpose method or technique can be used with different types of problems.
Perception is how the scene or environment is viewed. Analysis can be complicated as the perception varies based on the view angle, background etc. Today, AI has advanced sufficiently to be able to use Perception in multiple ways. Face Recognition and autonomous cars are a couple of examples of the usage of this component.
This AI component is a relatively difficult one as one has to take into account symbols and signs as well as the linguistic intonations of any language. A very important characteristic of human languages is its productivity and evolution, and such languages, as we know, can produce unlimited variety of sentences. This makes it a very difficult process to create an algorithm which can evolve as the language evolves. Another aspect of a language is the understanding. For a response or reaction, the language has to be understood.
Finally before we conclude, let us list a few examples of AI.
- Self – driving cars
- Digital assistants and smart speakers like Siri, Alexa etc.
- Facial Recognition
- X-Ray and MRI analysis etc. in Medicine
In conclusion, what is the Future of AI?
This field is one of the steadily growing ones with innovations and improvements happening at a steady pace. No one can predict the direction AI will take. But one thing is certain, AI is here to stay for a very long time.