What, why, & how of Artificial Intelligence
Artificial intelligence. To know understanding artificial intelligence is, we must know what intelligence is. Intelligence is the ability to acquire and apply knowledge and skills. This ability is what makes the difference between Humans, Animals, and machines & robots. If we make machines or robots to acquire and apply knowledge & skills and it is what called artificial intelligence.
It is a concept inspired by the working of our brain, and researchers have been researching and experimenting to make machines think and learn. We only know a little about working of our brain, we are trying to make artificial brain and this is why the journey of making artificial intelligence is really problematic and challenging. This is now in acceleration mode with all the Tech giants in this journey.
If you think that artificial intelligence is something that robots taking control over humans and destroying the world like the way we have seen in science-fiction movies, yes, it is artificial intelligence but at its worst-best, but it does not mean that it is the only artificial intelligence, it is also artificial intelligence.
What if I say we all have used AI many times? Excited! Shocked! Knowingly or unknowingly we are using it. Siri, Google Now and Cortana, Assisted Parking, smart home devices, and many other services. According to Wired, Fox, the AP, and Yahoo! All use AI to write simple articles like sports recap, financial summaries, etc. No, this article is not written by AI, but it is on AI.
AI is not a magic, it is just a code which is N times complex. It is a science or a method focused on designing intelligent systems & machines, using algorithmic techniques. Modern artificial systems use an artificial neural network, a computer code that emulates large networks of very simple interconnected units, similar to neurons in our brain. Suppose you have seen a car, not a car, the Lamborghini Aventador and the next time you see it, you will come to know that it is Lamborghini, but how? The neurons in brain checks with all the large network of images and then matches with Lamborghini and thus tells that it is Lamborghini and this is what exactly happens with an artificial neural network in machines. These artificial neural networks can learn from experience same as the neural networks in our brain. All of this happens through a set of coded programs designed to run neural networks with millions of units & billions of connections at blinding speed. This interaction with a large number of simple elements gives intelligence.
The modern artificial neural networks are good at recognizing patterns, translate the language, learning simple logical reasoning. AI is very good at recognizing patterns in a large amount of data which is not an easy task for humans.
How a machine become intelligent?
It is just like a kid learning from adults and doing mistakes & experiments. There are different ways of making a machine intelligence based on how they get intelligence. They have supervised learning, reinforcement learning, and predictive learning.
Supervised learning—we supervise and tell what correct answer is for a particular input. For a particular image input of the elephant, we tell the machine that it is an elephant and the correct answer is an elephant. A machine makes the prediction based on previous examples. This learning technique is the most common for training neural networks.
Predictive learning— the machine goes on telling answers either right or wrong. It goes on predicting and thus learns on its own.
Reinforcement learning— it is inspired by behaviorist psychology. For a particular input, the machine picks an action or sequence of action and gets a reward. This learning technique requires an extremely large number of trials to learn simple tasks. This technique is used when teaching a machine to play games like Chess, go or other simple video games.
For us, an image is an image but for machines & computers, images are an array of numbers. For a computer to recognize the image many layers of units are involved in it. In the very first layer machine recognize the edges of an object. I’n the next layers it recognizes simple shapes and in the next layer, it detects a combination of shapes which makes objects like a hand, a leg or the bonnet of a car. ‘In’ the final layer, it detects a combination of parts that form objects such as a car, an elephant, a person, etc. This object detection intelligence is very important in smart cars.
Artificial intelligence is not magic but it can create a lot of magical inventions. AI is serving millions every day and is making a huge difference in our lives. It is everywhere but it is invisible because it is just an algorithm.