You may have heard about the computer programme that technology company Microsoft recently launched that was able to recognize the breed of a dog from a photo. Of course there were many mistakes, but Microsoft was proud enough of the software to actually offer it to the public as it shows real success in the field of Artificial Intelligence (AI).
As the name tells you, AI is about the intelligence inside a machine. That machine could be a computer, mobile phone, a television or a refrigerator. The goal of AI is to make machines think like humans. To get to that stage, computer programmes adopt several strategies (plans) such as neural networks, machine learning and deep learning.
Neural networks are a way of building a computer such that it resembles a brain. Lots of different parts are wired together. While each part can work on simple questions (such as those that require a Yes or No answer), together, the whole can take on and answer complex or difficult questions. AI computer programmes are usually run on computers that have been built on the neural network design.
Machine learning is about teaching a computer by showing it lots of examples. Say, you want the computer to know what a traffic light looks like. Instead of exactly describing it (it has three colours, stands on a pole etc.), you would show it lots of pictures of traffic lights so that over time, the computer can make out a traffic light from a street lamp. It is through machine learning that Google is teaching its driverless cars-these cars are taken out on the roads in order to teach them to drive and they learn from their experiences on the road.
In deep learning, learning takes place at different levels. Say you want the computer to recognize a Christmas tree, you would feed it a lot of information at different levels – the shape of the tree, the kind of needle-shaped leaves it has, how it looks in different seasons and the kind of places it grows in. when looking at a picture of a tree, the computer then uses all this information which is in different sections and puts it together to make a decision if the tree is a Christmas tree or not. Each viewing experience is used by the computer to add to its knowledge – so, Deep Learning is a process that takes plenty of time and it uses a lot of data which has to be fed to the computer.
Examples of AI are already around us-it in computer programmes that can speak and driverless cars. Over time, the technology will get better and better-and then we can start worrying if machines will defeat humans and take over the world!