Moving in to AI

What is AI?

My search for the definition of AI led me to this:

"the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings" But how can a computer be Intelligent? Are we faking ourselves? Though AI is becoming more popular around the world, many philosophers disagree with the concept of a computer being intelligent. The "Chinese Room" contained a list of all the famous debates. The result of the Chinese Room Test is that computers cannot understand but can only manipulate based on programmed information.

But how do people believe that Computer is Intelligent? There comes more way of Approaches like the Turing Test, The Cognitive Modeling Approach, The Law of Thought Approach, and The Rational Agent Approach.

In Turing Test, AI should hold a conversation with humans. But how is this even possible? Here, NLP enters the picture. In this post, we'll dive deep into Turing Test and NLP.

How do Computers understand Humans?

You've probably heard of Sophia, the Humanoid Robot. She can speak, act, and even think like a human at times. Have you ever thought about how Sophia reacts appropriately based on the conversation? All of this is the result of NLP. Humans interact with and understand each other through the use of words and writing. But how does a computer evaluate data that is only 0's and 1's? NLP, a branch of AI, assists machines in reading, understanding, and deriving meaning from human languages.

Pipelines of NLP

The idea behind NLP is to train computers to learn and understand languages using appropriate Algorithms.

There are seven steps to performing NLP.

It's somewhat similar to teaching a kid a language. The first among these is segmentation. When teaching a kid a new language, we divide a paragraph into separate sentences for better understanding. That is what segmentation is all about. Tokenization is the next step. Following the division of the paragraph into Sentences, our next goal is to divide each Sentence into separate words. Stemming: The Kid does not know what the separated term is simply by separating the words. As if it is a verb, noun, or something. So we utilize stemming for this. Stemming predicts the parts of speech for each divided word. Consider the word 'playing'. Lemmatization is the process of removing the end i.e. ing and obtaining the root stem.

When I first started learning a new language, articles and other non-essential words annoyed me. So I began removing non-essential terms and focused more on the key action to be done. As a result, I was able to figure out the sentence's meaning. The same holds for stop word analysis.

So now each paragraph has been divided into sentences, and each sentence has been divided into words. How do we connect those? Through the use of Dependency Parsing. It focuses on analyzing and connecting the words.

So, the process is now complete, and the final stage is Parts of Speech Tagging. POS contains all of the verbs, adverbs, and other words that assist us understand the meaning of the sentence. Thats' how NLP works.

Turing Test

Back to Turing Test, Is it possible for a computer to think like Humans? Alan Turing, The Founder of the Turing Test, stated that a computer is said to be intelligent if it can respond or act like a Human. So how it works? So, let me give you a simple example that we all hate. When we try to log in to the website or access new pages, we get a page that confirms that we are human. The same principle works here as well. Consider two humans and a computer. One human is a player, another is a judge, and the computer is also a player. The Judge is separated from both Players. But he is aware of the players. The questioner asks a series of questions to both the human and the computer. Following a series of questions, the questioner determines whether the answers were provided by a human or a computer. If the questioner sees the computer's answers as human responses, the computer is said to be intelligent. Because it can think like a human. So, now we're thinking about what kinds of questions we can ask.

There are no specific questions that need to be asked while performing Turing. However, the suggested method is to ask questions on several aspects of human intelligence and behavior.

From "Can Computers be Intelligent" to "Will Computers Replace Humans?", it's clear that no Computers can replace Humans but rather act according to the given Instructions.