AI Explained : Meaning, Brief History and Roadmap

Use of Artificial Intelligence (AI) has been accelerating across all industries in the recent years leading to increased business as well as job opportunities and making AI a very hot topic for discussions. We have all heard that AI is the future but probably the time has now come where AI is not the future but something happening today. This decade would witness deployment of AI at massive scale and by 2030, there would be hardly anyone on planet who would not be leveraging AI to its benefit directly or indirectly.

Given the impact that AI is going to make, it is essential that every person understands what AI really is and how they could be impacted/ benefitted with AI. This is the first article of this Newsletter wherein we would be covering AI Concepts, News, Success and Failure Stories, Use-cases across various industries, Top Companies and many other things that you should be knowing.

Before we try to understand what Artificial Intelligence really is, let us first decode the real meaning of the term "Intelligence":

How Do We Define Intelligence?

Humans define intelligence as the ability to solve complex problems or make decisions in unforeseen and trying situations.The same can be said about machines. Traditionally, machines were not trained to do that but like everything else, they evolved as well.

In 1959, Alan Turing, an English Mathematician, devised the “Turing Test”. According to the Turing Test, a machine is considered "Intelligent" if it is difficult to distinguish it from a human owing to its behavior.

This breakthrough is considered to be one of the foundation stones of modern Artificial Intelligence.

Introduction to Artificial Intelligence

The meaning of Artificial Intelligence has evolved over the years. In today’s world, Artificial Intelligence can be defined as a machine’s ability to simulate/ mimic human-like thinking, problem-solving, and decision-making processes.

With Google's Search Engine to Netflix Recommendation System, Self Driving Cars as well as Alexa and Siri as real-world examples of AI in our daily life, the benchmark for AI performance is set in terms of reasoning, speech, and vision tasks.

There are a few common terms that you would usually come across when people talk about AI, such as Machine & Deep Learning, Internet-of-Things, Natural Language Processing, Fuzzy Logic, Computer Vision, Robotics, etc. We will get to these topics in further detail in a later segment.

First, let us understand how modern Artificial Intelligence evolved into what it is today by diving into the history of AI.

History of Modern AI

1956

Artificial Intelligence was first mentioned in a conference led by John McCarthy at Dartmouth University, in 1956. Due to this conference, two of the most fundamental approaches to AI were implemented. These were:

Top-Down Approach

This was a rule-based approach where the computer programs were programmed with the rules that guide human behavior. Commonly used modern machine learning algorithms such as Decision Trees fall under this approach.

Bottom-Up Approach

Some researchers favored a bottom-up approach which implied creating models that simulated or mimicked the functioning of the human brain. This planted the seed for what came to be known as Neural Networks.

1969

Marvin Minsky developed a general-purpose robot, named Shakey. Shakey could choose its actions based on the environment but it was extremely slow.

1988

IBM published “A Statistical Approach to Language Translation”

Machine Learning became based on a statistical analysis of past events and data — not on a deep understanding of the tasks to be done.

1997

IBM’s supercomputer, Deep Blue beat Gary Kasparov in a chess match held in Philadelphia. Deep Blue could analyze over 200 million moves per second and could think strategically.

2002

Artificial Intelligence enters homes in the form of Roomba, a vacuum cleaner.

2008

Apple implemented sophisticated speech recognition for iPhone for the first time.

2012

AlexNet could classify random images with an error of 16%.

2014

The first self-driving car by Google passed a state driving test.

2015

ResNet could do image classification with an error of 3.7% whereas humans had an error of 5%.

March 2016

Google’s AlphaGo defeated the world Go player champion, Lee Sedol.

For comparison, there are 20 possible first moves in chess. In Go, there are 361.

Evidently, Go is a lot more complicated than chess, which is proof that AI had already begun to surpass humans in certain tasks.

Machine Learning, DL & Data Science in AI

Let us now understand some of the discernable components briefly that entail or are entailed by modern-day Artificial Intelligence and how they work in tandem.

Machine Learning

These days machines can learn and adapt new patterns in decision-making without any explicit instructions. This is made possible by Machine Learning algorithms. The algorithms are trained on previously available data so that they can learn useful patterns from that and then put them to sophisticated prediction tasks.

Deep Learning

The Holy Grail in the search for modern artificial intelligence is figuring out to make a machine behave like a human. In other words, if an algorithm can mimic brain functions, it is more than likely to behave like a human. Deep Learning, in the form of Neural Networks, provides us with the ability to create increasingly complicated, deep, and diverse models that mimic the biological functioning of the brain, enabling machines to perform complicated tasks such as image and speech recognition, natural language processing and generation, stock market predictions, etc.

Internet-of-Things

IoT can be thought of as a web connecting multiple physical devices, with different functionalities, that can exchange data with one another over the internet. Keeping that in mind, it is quite evident that the integration of Artificial Intelligence with Internet-of-Things is imminent. While IoT deals with devices interacting using the internet, AI makes the devices learn from their data and experience.

Fuzzy Logic

The word “Fuzzy” usually refers to things that are unclear or vague. It is an approach to problem-solving based on the concept of “degrees of truth” rather than considering truth as just a binary characteristic.

Consider a bowl of water. Using traditional logic, whether it is hot or cold can be denoted by just two features. Hot and cold. With fuzzy logic, the temperature of the water can be represented on a spectrum or gradient ranging from very hot to very cold. Fuzzy logic enables us to get a much smoother output instead of discrete values such as 1 and 0.

Natural Language Processing

What is natural language processing? You are doing it right now. You are listening to or reading these words and sentences and forming some sort of comprehension from it to understand the context better. When we make a machine do that for us, it’s called Natural Language Processing. NLP enables the development of various products and services based on sentiment analysis and speech recognition among others.

Computer Vision

Computer vision is an area of artificial intelligence that allows machines to extract useful information from digital images, videos, and other visual inputs and take actions or make recommendations based on that information, much like human vision. To put it another way, if AI gives machines the ability to think, computer vision gives them the ability to see, observe, and comprehend.

Future of AI

So how does the future of AI look in the coming years?

• According to Gartner, AI usage has increased by 270% from 2015 to 2019.
• By 2025, legislation, according to Gartner, will need a focus on AI ethics, transparency, and privacy, which will drive, not inhibit, trust, growth, and better AI functioning globally.
• All the major tech companies of the world are focused on developing autonomous projects for their products.
• According to Warren Buffet, Chairman of Berkshire Hathaway, we might not be too far away from a future where all the goods and services we receive today might just be provided to us at the push of a button every morning by AI-powered systems.
• AI is slowly seeping into every aspect of our lives and if harnessed properly, it promises to create a vastly more productive and efficient economy.
• It is estimated that artificial intelligence has the potential to double every industry’s growth to date.
• AI has the ability to generate enormous prospects for people and make lives better in general.

With this, let me conclude this article by saying that irrespective of your educational background and occupation, you should definitely be trying to understand how AI could benefit you. In the upcoming articles of this series, we would be discussing the real life use-cases of AI across all industries.

Thank you for reading this. Signing Off !