Chapter 1: What is AI? (Part 1)

Defining Computers in Real Terms

We will first step away from sensationalized stories about robot uprisings in films to see what AI actually does in real life and how. And for that, we first need to understand what a computer is and how it works. A computer is really just a massive collection of switches that control the flow of electricity. Just as we put up a water barrier (naka) in our fields or irrigation beds (kyari) to stop or let water flow, transistors (or switches) in a computer work to stop or allow the flow of electricity. And it is in the form of different arrangements of these transistors (Logical Gates) that human language and logic rules are fed into computers. For example, if electricity is passing through the very first transistor while all others are closed, we can feed (or code) this arrangement as 'yes' or 'no' (1 and 0). And by combining billions of such transistors and rules, we build 'Artificial Intelligence' (AI) that works like the human brain. In this booklet, we will see how AI today drives many everyday technologies, such as cars finding their way automatically (Autonomous Navigation), finding everything according to preference on online websites (Recommendation Engines), and photo and face recognition systems (Visual Recognition Systems). To understand all of these well, we need to understand two fundamental concepts:

• Autonomy: This means any AI system's ability to work independently without human intervention at every step in a changing and complex environment. Understand this through the role of a traditional Guval (herdsman) in Marwar, who takes his herd of livestock to graze in an Adavo or Oran. Along the way, he faces many challenges, such as changing paths, crossing a small stream, a sheep getting left behind, or getting all those sheep across a road, and so on. The Guval does not instruct every sheep or camel on every single step; instead, the entire herd navigates and keeps moving autonomously within set boundaries. Similarly, an autonomous software agent evaluates its environment and makes its own decisions. Although comparing a software agent to any living creature is not strictly fair, to explain certain concepts we sometimes have to use such rudimentary analogies. If you have a better example, please share it with us.
• Adaptivity: In our traditional culture, teachings received from ancestors and knowledge gained through experience continue to pass down from generation to generation. A young farmer learns from the experience of many seasons when to harvest Phali (Khejri pods) or how to estimate the arrival of a Sunto (monsoon storm) by looking at the sky. Gaining diverse experiences within the same environment (i.e., weather/seasons) is called adaptivity in the computer world. This is precisely the difference between software with fixed rules and state-of-the-art AI systems driven by machine learning models. Because AI systems automatically improve when they receive new data.

Mapping the Computing Landscape

To understand this subject well, students should have a clear map of how AI connects with other branches of science. If we attempt to connect these branches to one another or place them inside one another, we can say that Computer Science is the outermost large boundary, inside of which sits Artificial Intelligence. Inside that comes the sphere of Machine Learning, and even deeper within it, Deep Learning can be placed.

Data Science is an interdisciplinary field that combines Machine Learning, statistics, and domain expertise to analyze data and uncover patterns related to the past, present, and future. Meanwhile, Robotics focuses on designing and building physical machines equipped with sensors and actuators that interact with and operate within the real physical world.