Demystifying AI: #1

As we have reached the end of our very successful ‘AI lecture series’ at Hacker Dojo, I would like to take this opportunity to provide a foundation for beginners to start their AI journey. When I started my own journey, AI and Machine learning were distant technologies that I couldn’t be bothered to learn about, as they didn’t affect my day to day work or existence – or so I thought. Boy, was I wrong! Even if you are unable to define Artificial Intelligence or believe that AI is all science fiction, the fact is - AI is very much here. Almost every one, including my retired father, is using it without them even realizing it. AI has already been silently infused into the products and services that we all use on a regular basis.

Do you use email? Well, you are already using AI via the spam filtering feature of your email provider. Do you buy on Amazon? The recommendations regarding what products you should buy are powered by AI. You like Amazon's quick delivery? Well, the Kiva robots in Amazon's warehouses enable that quick delivery. What about entertainment? Netflix, Spotify? The movie and song recommendations are powered by AI. Are you on Facebook? The face recognition feature is enabled via AI. Alexa, Siri? Again, it is all powered by AI. As you can see, AI is already integral to the many services that you use on a daily basis and now that usage is going to further explode to almost all industries, products and services. A Professor from Stanford University, and a thought leader in AI, Andrew Ng, has already proclaimed – AI is the new electricity. Just as electricity transformed everything as people knew then, AI is going to transform everything as we know now.

The first lecture in the Hacker Dojo AI lecture series delivered by yours truly was an overview of AI for beginners. The lecture provided a mental model, a framework for beginners by which they could start understanding the various technologies within AI. Differentiating between related technologies and use cases, instead of using them interchangeably is key to understanding AI in the most efficient way.

So, let’s dive into it - What is a very simplistic definition of AI?

AI is a set of technologies that enable machines to perceive, learn, think and act, just like humans do.

ARTIFICIAL INTELLIGENCE (AI)

As you will notice from the above figure - Computer Vision and NLP (natural language processing) are the technologies that enable computer/machines to perceive things whereas Machine learning enables computers to learn, think, reason and Autonomous Vehicle, Robotics, Chatbots and Virtual assistant are action enabling technologies. We will be covering all these things in subsequent articles. But, lets switch to the two terms that people use interchangeably – AI and Machine learning.

What is Machine learning?

Machine learning is a subset of AI. Machine learning is the science of getting computers to learn, from the data presented to them, from experiences, without being explicitly programmed. Machine learning is one approach of AI, an approach that is based on statistics; hence it is also called the statistical approach or probabilistic approach.

So, If ML is one approach, what is the other approach? The other approach is the deterministic approach or rules based approach. In Deterministic or rules based approach, machines are programmed directly by the experts with a company’s best practices in mind. The early part of the AI history was all about the deterministic approach. The rise of expert systems in the early 1980’s was a direct result of this deterministic approach being at the forefront of AI development. However, the current explosion of AI is due to advances in Machine Learning but primarily Deep Learning.

So, then, what is Deep Learning?      

Deep learning is a type of Machine Learning that consists of inter - connected layers of ‘neurons’ (software based calculators) that form a ‘neural network’. The network ingests more data, processes data through multiple layers, learns complex features of data, uses what it has learned to make determinations about new data and provide more accurate results. Here is the timeline and the hierarchy of AI/ML/DL.

Source: Nvidia

With above frameworks/mental models as a starting point, we will dig deeper into Machine learning and Deep Learning in the future articles.