AI 1.01 — Separating facts from fiction

mediumThis post was originally published by Ram Vignesh Namasivayam at Medium [AI]

Artificial Intelligence: A General Overview

No matter which stage of your career you are in, this article will give you an overview of what to expect from Artificial Intelligence.

Is AI a General Purpose Technology?

General Purpose Technologies like Steam Engine in the 18th century, Electricity from the 19th century, computers and the Internet from the 20th century have accelerated Innovations, created new Opportunities, and was a cause for economical growth.

AI is the New Electricity

Just as electricity transformed almost everything 100 years ago, today I actually have a hard time thinking of an industry that I don’t think AI will transform in the next several years

– Dr. Andrew Ng

In the 21st century, it is Artificial Intelligence, particularly Machine Learning, is the most important general-purpose technology of our era.

The evolution of computers over a period of time can be distinguished with a major technological advancement that has influenced the purpose of computers. There were vacuum tubes, transistors, IC’s, and now AI. Though AI is Significant and already has many use cases there are many misconceptions about AI and has generated unrealistic expectations around the world. Some see it as a disruptive technology some hype it as a cure-all and there is debate on whether it is a job creator or destroyer.

The Buzz word

I got introduced to the term AI through science fiction movies like Imitation Game, Her, and Iron man. Fascinated by the idea of Jarvis, I explored conversational AI bots. we have voice assistants like Google Assistant, Siri, and Alexa but do they interact the same way as in movies? Not Exactly. Voice bots interact with humans to answer questions and display vocalized keywords. these assist users by just recognizing the voice whereas we have seen Jarvis helping out Ironman in critical Decision Making and in Her we have seen AI help the protagonist, an introverted writer, help him write letters. Moreover, when he finds out about the AI’s ability to learn and adapt, he even falls in love with it. It is Fictional right? So, It may or may not happen in real life. Is there an AI solution that can assist us in creating music, art, and programming that is portrayed in movies?

Hype Cycle

The Hype cycle from Gartner highlights the expectation levels of AI technologies that could create a significant impact on industry, society, and business. In 2019, it was predicted Speech Recognition is less than two years to mainstream adoption to deliver significant results. Some technologies will be dropped if they do not contribute significantly as per expectations of market demand.

AI is starting to deliver on its potential and its benefits for businesses are becoming a reality

What Can AI do as of now?

AI has already provided many solutions and we may not even realize it. Even Alexa, Google Assistant uses AI algorithms. Youtube’s Recommendation engine uses AI. There are major advancements in perception and cognition. There are Models that can surpass human-level performance in imagenet classification and we have seen AI beat humans in many games. These Advancements are due to High Computing power, enormous amounts of data and are mostly in solving problems related to labeled data.

Generative AI can generate new content like text and images that preserve some relationship to previously learned representations of training data. Open AI’s GPT-3, a language model that uses deep learning produces human-like text and there are Generative Adversarial Networks that can generate even Modern Art.

How is Machine Learning different from programming?

Artificial Intelligence is a branch of computer science dealing with the simulation of human behavior in computers.

Machines trying to mimic human behavior.

My machine can do tasks for which I programmed for. How do machines understand Visual Perception, Recognizing speech, Identify differences between a dog and a cat?

Machine Learning is a form of AI that enables a system to learn from data rather than through Explicit Programming to improve its performance at a given task

Machine’s Ability to keep improving the performance without humans having to explain exactly how to accomplish a given task

Information technology has focused predominantly on programming machines through procedures by transferring knowledge from what we know and understand to execute a specific task, for almost 50 years now. the flaw in the approach is that we cannot write programs for tasks that we cannot fully understand. To overcome this flaw we use machine learning to solve problems by learning from examples through a feedback

Various AI Approaches

Though Machine Learning is a vital part of AI, there are multiple methods like statistical machine learning, neural networks, deep neural networks, and so on. Semantics for Natural language processing tasks, and Logic-based rule engines for simple decision making. Deep Learning excels at recognizing images and speech. Unsupervised learning systems seek to learn on their own.

Humans are excellent unsupervised learners. we pick most of the knowledge of the world with little or no labeled data. But, it is exceedingly difficult to create models that can learn without labeled data.

Now we have understood that we don’t always have to program the machines. they can Learn by Themselves. But it still has a long way to go. So does that mean machines think? Someone has raised such a question about seven decades ago. Guess who? follow this article to learn about “ Can Machines Think ? ”

When we start reading about Artificial Intelligence we hear a lot of arcane terms like reinforcement learning, neural networks, GAN’s, Computer vision and lack a structured learning approach which in turn complicates the understanding. I am trying to simplify these jargons and demystify misconceptions about AI in a series of Articles.

Spread the word

This post was originally published by Ram Vignesh Namasivayam at Medium [AI]

Related posts