Before we dive deep into Artificial Intelligence (AI) and Machine Learning (ML), it’s important to understand what intelligence really means and how we can measure it.

To me, intelligence is a combination of multiple capabilities:

  • The ability to observe and gather inputs through one or multiple modes (touch, smell, vision, hearing, speech, intuition) using our central processor — the brain
  • Memory: storing information for future use
  • Converting information into useful knowledge
  • Applying that knowledge through actions — for survival, productivity, creativity, pleasure, or even destruction

Every species and every individual exhibits these abilities at different levels. When it comes to machines, our goal is to enable them to think and act like humans.

This is what the field of Artificial Intelligence aims to achieve.

AI includes several important subfields, such as:

  • Machine Learning (ML)
  • Deep Learning (DL)
  • Natural Language Processing (NLP)
  • Computer Vision (CV)
  • Speech-to-Text and Text-to-Speech technologies

Among many definitions of AI, one which appeals to me is from the book Grokking Machine Learning: The set of all tasks in which computers can make decisions by:

  • Mimicking the ways a human makes decisions
  • Mimicking genetic processes / evolutionary processes / physical processes

Any time we see a computer solving a problem by itself, be it driving a car, finding a route between two points, or recommending a movie, we are looking at AI.

Machine Learning (ML) is a part of AI and can be defined as a set of all tasks in which a computer can make decisions based on data.

Let’s take an example where we want to buy a car. There are two ways to decide:

  1. We carefully look at the features such as price, mileage, navigation & try to figure out the best combo that adjusts to our budget. That is using “logic & reasoning”.
  2. We talk to our connections/network/friends/family about which car they own, what is their feedback. We form a list of information collected & use that list to decide. So here we are using “experience”.

ML represents the second method: “Making decisions using our experience (data in computer lingo)”.

This is a major shift where rather than using any means necessary to solve a problem, we use only data. To understand it, let’s take an example and compare it with our traditional software engineering.

Let’s say the problem statement is to identify if an image contains an apple. If we start writing a program to develop this task, we will find it to be extremely complex. Let’s ponder on how we humans learned to recognize an apple? We learned it by repetition. We saw many objects in our childhood & adults would tell us what these objects were.

To learn what an apple was, we saw many apples throughout the growing period while hearing the word apple, until one day it clicked & we knew what an apple was.

On similar lines, we show the computer many images & we tell it which ones contain an apple. We repeat this process until the computer catches the right patterns & attributes that constitute an apple.

At the end of the process, when we feed the computer a new image, it can use these patterns to determine whether the image contains an apple. Of course, we still need to program the computer so that it catches these patterns.

Updated: