A lot is heard and talked about ‘Artificial Intelligence’, popularly called as AI, mainly because it is helping humans to solve complex problems. However, for a layman who is not working in the technology space, AI sounds like a complex topic, something that is difficult or impossible to understand. So, here is an attempt to explain it in a very simple way.
AI is a branch of computer science. It is about asking machines to perform tasks that typically require human intelligence.
Let’s think of the human brain for a moment.
The human brain is capable of Learning. It can learn from past experiences or observations. A simple example could be estimating travel time to work based on past experience.
The reasoning is another capability that the human brain has. It can understand patterns, it can co-relate events or it can make logical conclusions based on the information it has stored. Like anticipating rain when there is cloudy weather.
The human brain is also capable of perceiving things. Like identifying objects – a cat, a car, or a specific brand based on its logo.
So, through AI, we are simulating the human brain in a computer, by writing instructions which are nothing but computer programs.
AI is a broad term. It has multiple subsets within it.
- Machine Learning,
- Deep Learning also called Artificial Neural network,
- Natural Language Processing,
- Computer Vision,
- Robotics, and
- Expert Systems
Let’s start with Machine Learning.
MACHINE LEARNING
Machines can learn in two ways as supervised learning and unsupervised learning.
In supervised learning, machines are trained and tested on past data. For example, we can train a machine to forecast water demand for a town based on certain inputs like population, weekday or weekend, time of the day, etc from past data.
In machine learning terms, the water demand is called the ‘Label’, while the inputs as population, weekday or weekend, etc are called ‘Features’.
In the case of unsupervised learning, the machine is not trained because of the absence of labels. The algorithm allows the machine to act without guidance. It is used to group unstructured data, based on similarities and distinct patterns in the data. A simple example could be segmenting utility customers as defaulters or good customers who pay their bills on time.
DEEP LEARNING
‘Deep learning’ is another subset of machine learning. It’s also called ‘Artificial Neural Networks’ aka ‘ANN’. The working principle of ANN is the same as the human brain, where neurons are connected in layers and they pass information to the next layer. Same way, in Neural Networks, information is passed through multiple layers of connected nodes to perform certain tasks.
ANN can help Identify leakages in water distribution networks using data from various sensors.
NATURAL LANGUAGE PROCESSING
Another mostly used subset of AI is ‘Natural Language Processing’ aka ‘NLP’. In NLP, machines are trained to read, decipher, and make sense of human languages.
There are a variety of applications of NLP. It can help classify sentiments of customers as positive or negative by reading their social media posts. NLP can translate words from one language to another like Google translator. It can also be used to get answers from machines by asking them questions in human language, like, a chat-bot on a website to respond to customer queries or SIRI in Apple devices.
COMPUTER VISION
‘Computer Vision’ is another set of AI, in which objects from videos and images can be identified using Artificial Neural Networks. For example, using this technique, the machine can analyze objects in a video captured by a CCTV camera. The machine can then be programmed to send an alert if any suspicious activity or an object is identified by the machine.
ROBOTICS
There may be different opinions about ROBOTICS whether it is a field of AI actually. In my view, it’s a field of engineering. It overlaps with mechanical, electronics, nanotechnology, and AI.
As a simple example of Robotics, imagine a robot of the size of a small bug that can run through a pipe network to assess the condition of the assets.
EXPERT SYSTEMS
There is one more field of AI called Expert Systems’ which is simulating machines to act as subject matter experts. For example, mimicking the intelligence of a cancer surgeon. This field of expert systems is maturing and looks promising to solve many complex problems as more and more data is being collected.
This was a quick tour of what AI is, what it includes with some of its applications. Hope you found this article useful. Thank you for reading.